Datasets:
Add files using upload-large-folder tool
Browse filesThis view is limited to 50 files because it contains too many changes. See raw diff
- .DS_Store +0 -0
- .gitattributes +39 -0
- AI_powered_CCRM_liver.pdf +3 -0
- AI_powered_CCRM_liver__reading_notes.md +180 -0
- README.md +281 -0
- README.md.bak +281 -0
- STRUCTURE_REPORT.md +245 -0
- _audit_report.md +55 -0
- _provenance/uuid_to_new_name.csv +24 -0
- _view_only_list.md +71 -0
- merged_views/MERGE_REPORT.md +190 -0
- merged_views/dashboard.html +0 -0
- preview.png +3 -0
- sc__hcc-cd45-ss2__sharma2020-ss2-6pts__7074cx54574g__GSE140228-ss2.h5ad +3 -0
- sc__hcc-cd45__guo2025-9pts__191435cx28671g__GSE235863-9pt.h5ad +3 -0
- sc__hcc-cd8tcell__guo2025-5pts__95408cx26256g__GSE235863-5pt.h5ad +3 -0
- sc__hcc-fetal__sharma2020-62pts__109238cx2384g__GSE156625-HCCF.h5ad +3 -0
- sc__hcc-hbv-caf__zhang2023-11pts__115732cx36601g__GSE202642.h5ad +3 -0
- sc__hcc-iccA-mixed__ma2019-set1-12pts__5115cx20124g__GSE125449-set1.h5ad +3 -0
- sc__hcc-iccA-treated__ma2021-46pts__56721cx18667g__GSE151530.h5ad +3 -0
- sc__hcc-mash-spectrum__huang2025-4pts__34396cx32285g__GSE282630.h5ad +3 -0
- sc__hcc-multisite__lu2022-10pts__71915cx25712g__GSE149614.h5ad +3 -0
- sc__hcc-tcell__zheng2017-6pts__5063cx23459g__GSE98638.h5ad +3 -0
- sc__hcc-trm__park2025-7pts__41848cx36601g__GSE281110.h5ad +3 -0
- sc__healthy-nat__liver__13083cx33694g__GSM4648565.h5ad +3 -0
- sc__healthy__b-cell__1250cx32596g__6ade4ff5-368c-4276-b051-818dc954da6d.h5ad +3 -0
- sc__healthy__cholangiocyte__1011cx32596g__601ef580-74ce-4a87-96e8-8e22bf4ed9fa.h5ad +3 -0
- sc__healthy__endothelial__9422cx32596g__a95e1659-9a48-4c55-8062-621ee4df9160.h5ad +3 -0
- sc__healthy__hepatocyte-v1__53015cx32596g__63137dcf-6236-464d-9018-e58c9323f59c.h5ad +3 -0
- sc__healthy__hepatocyte-v2__13635cx32596g__6b241dde-25c0-4edc-9021-03603ec3a524.h5ad +3 -0
- sc__healthy__lymphoid__16665cx32596g__9bc7506b-7e73-4cea-bbb3-3603a016fbca.h5ad +3 -0
- sc__healthy__macrophage__11127cx32596g__2cfec927-9163-4684-ae04-c15175a6d781.h5ad +3 -0
- sc__healthy__stellate__1417cx32596g__08a9f031-1e76-405a-8050-0635743ce187.h5ad +3 -0
- sc__psc-pbc-healthy__all-cells__89637cx32596g__7d4d0da4-655e-438a-a2ec-b4371e2b80fc.h5ad +3 -0
- sn__hcc-tumor-normal__alvarez2022-1pts__39995cx58100g__GSE189175.h5ad +3 -0
- sn__hcc-tumor-normal__alvarez2022-3pts__39995cx58100g__GSE189175.h5ad +3 -0
- sn__psc-pbc-healthy__all-cells__105780cx32596g__4b5895d7-6d92-471a-b13a-5c59a000ddc4.h5ad +3 -0
- visium__hcc-antiPD1__liu2025-1pts__1320cx36601g__GSE238264-HCC5NR.h5ad +3 -0
- visium__hcc-antiPD1__liu2025-1pts__2170cx36601g__GSE238264-HCC3R.h5ad +3 -0
- visium__hcc-antiPD1__liu2025-1pts__2453cx36601g__GSE238264-HCC7NR.h5ad +3 -0
- visium__hcc-antiPD1__liu2025-1pts__2575cx36601g__GSE238264-HCC6NR.h5ad +3 -0
- visium__hcc-antiPD1__liu2025-1pts__2766cx36601g__GSE238264-HCC2R.h5ad +3 -0
- visium__hcc-antiPD1__liu2025-1pts__3002cx36601g__GSE238264-HCC4R.h5ad +3 -0
- visium__hcc-antiPD1__liu2025-1pts__3006cx36601g__GSE238264-HCC1R.h5ad +3 -0
- visium__healthy-C73__blockA1__4992sx35477g__9c731db0-9f44-4f8c-8139-9b9af2bcc782.h5ad +3 -0
- visium__healthy-C73__blockC1__4992sx35477g__b2287ef1-eac3-49cc-93de-65df74e26a61.h5ad +3 -0
- visium__healthy-C73__blockD1__4992sx35477g__44453de7-1d66-4bd1-a83d-0b73a1690d57.h5ad +3 -0
- visium__psc-PSC011__blockA1__4992sx35477g__7bc39166-b664-4c92-922d-f9a8047768d2.h5ad +3 -0
- visium__psc-PSC011__blockB1__4992sx35477g__6da751d4-63af-43b8-96db-395ca73dfb5f.h5ad +3 -0
- visium__psc-PSC011__blockC1__4992sx35477g__0cc59004-2b35-4767-8278-83e097ef32d1.h5ad +3 -0
.DS_Store
ADDED
|
Binary file (14.3 kB). View file
|
|
|
.gitattributes
CHANGED
|
@@ -58,3 +58,42 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
|
| 58 |
# Video files - compressed
|
| 59 |
*.mp4 filter=lfs diff=lfs merge=lfs -text
|
| 60 |
*.webm filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 58 |
# Video files - compressed
|
| 59 |
*.mp4 filter=lfs diff=lfs merge=lfs -text
|
| 60 |
*.webm filter=lfs diff=lfs merge=lfs -text
|
| 61 |
+
sc__hcc-fetal__sharma2020-62pts__109238cx2384g__GSE156625-HCCF.h5ad filter=lfs diff=lfs merge=lfs -text
|
| 62 |
+
visium__psc-PSC011__blockC1__4992sx35477g__0cc59004-2b35-4767-8278-83e097ef32d1.h5ad filter=lfs diff=lfs merge=lfs -text
|
| 63 |
+
sc__healthy__lymphoid__16665cx32596g__9bc7506b-7e73-4cea-bbb3-3603a016fbca.h5ad filter=lfs diff=lfs merge=lfs -text
|
| 64 |
+
sc__psc-pbc-healthy__all-cells__89637cx32596g__7d4d0da4-655e-438a-a2ec-b4371e2b80fc.h5ad filter=lfs diff=lfs merge=lfs -text
|
| 65 |
+
sc__hcc-iccA-treated__ma2021-46pts__56721cx18667g__GSE151530.h5ad filter=lfs diff=lfs merge=lfs -text
|
| 66 |
+
AI_powered_CCRM_liver.pdf filter=lfs diff=lfs merge=lfs -text
|
| 67 |
+
visium__psc-PSC011__blockB1__4992sx35477g__6da751d4-63af-43b8-96db-395ca73dfb5f.h5ad filter=lfs diff=lfs merge=lfs -text
|
| 68 |
+
sc__healthy__b-cell__1250cx32596g__6ade4ff5-368c-4276-b051-818dc954da6d.h5ad filter=lfs diff=lfs merge=lfs -text
|
| 69 |
+
sc__hcc-multisite__lu2022-10pts__71915cx25712g__GSE149614.h5ad filter=lfs diff=lfs merge=lfs -text
|
| 70 |
+
sn__psc-pbc-healthy__all-cells__105780cx32596g__4b5895d7-6d92-471a-b13a-5c59a000ddc4.h5ad filter=lfs diff=lfs merge=lfs -text
|
| 71 |
+
visium__hcc-antiPD1__liu2025-1pts__3002cx36601g__GSE238264-HCC4R.h5ad filter=lfs diff=lfs merge=lfs -text
|
| 72 |
+
sc__hcc-iccA-mixed__ma2019-set1-12pts__5115cx20124g__GSE125449-set1.h5ad filter=lfs diff=lfs merge=lfs -text
|
| 73 |
+
sc__hcc-cd45__guo2025-9pts__191435cx28671g__GSE235863-9pt.h5ad filter=lfs diff=lfs merge=lfs -text
|
| 74 |
+
sc__hcc-hbv-caf__zhang2023-11pts__115732cx36601g__GSE202642.h5ad filter=lfs diff=lfs merge=lfs -text
|
| 75 |
+
visium__hcc-antiPD1__liu2025-1pts__2575cx36601g__GSE238264-HCC6NR.h5ad filter=lfs diff=lfs merge=lfs -text
|
| 76 |
+
visium__healthy-C73__blockA1__4992sx35477g__9c731db0-9f44-4f8c-8139-9b9af2bcc782.h5ad filter=lfs diff=lfs merge=lfs -text
|
| 77 |
+
visium__healthy-C73__blockD1__4992sx35477g__44453de7-1d66-4bd1-a83d-0b73a1690d57.h5ad filter=lfs diff=lfs merge=lfs -text
|
| 78 |
+
sc__healthy__macrophage__11127cx32596g__2cfec927-9163-4684-ae04-c15175a6d781.h5ad filter=lfs diff=lfs merge=lfs -text
|
| 79 |
+
sc__hcc-cd8tcell__guo2025-5pts__95408cx26256g__GSE235863-5pt.h5ad filter=lfs diff=lfs merge=lfs -text
|
| 80 |
+
sc__healthy__stellate__1417cx32596g__08a9f031-1e76-405a-8050-0635743ce187.h5ad filter=lfs diff=lfs merge=lfs -text
|
| 81 |
+
visium__hcc-antiPD1__liu2025-1pts__1320cx36601g__GSE238264-HCC5NR.h5ad filter=lfs diff=lfs merge=lfs -text
|
| 82 |
+
visium__psc-PSC011__blockD1__4992sx35477g__721bfc1c-f77f-4c5a-afc6-04e3e3c675d3.h5ad filter=lfs diff=lfs merge=lfs -text
|
| 83 |
+
sc__hcc-trm__park2025-7pts__41848cx36601g__GSE281110.h5ad filter=lfs diff=lfs merge=lfs -text
|
| 84 |
+
sc__hcc-cd45-ss2__sharma2020-ss2-6pts__7074cx54574g__GSE140228-ss2.h5ad filter=lfs diff=lfs merge=lfs -text
|
| 85 |
+
visium__hcc-antiPD1__liu2025-1pts__2170cx36601g__GSE238264-HCC3R.h5ad filter=lfs diff=lfs merge=lfs -text
|
| 86 |
+
sc__healthy__endothelial__9422cx32596g__a95e1659-9a48-4c55-8062-621ee4df9160.h5ad filter=lfs diff=lfs merge=lfs -text
|
| 87 |
+
sc__healthy__cholangiocyte__1011cx32596g__601ef580-74ce-4a87-96e8-8e22bf4ed9fa.h5ad filter=lfs diff=lfs merge=lfs -text
|
| 88 |
+
sc__healthy__hepatocyte-v1__53015cx32596g__63137dcf-6236-464d-9018-e58c9323f59c.h5ad filter=lfs diff=lfs merge=lfs -text
|
| 89 |
+
visium__hcc-antiPD1__liu2025-1pts__2766cx36601g__GSE238264-HCC2R.h5ad filter=lfs diff=lfs merge=lfs -text
|
| 90 |
+
visium__psc-PSC011__blockA1__4992sx35477g__7bc39166-b664-4c92-922d-f9a8047768d2.h5ad filter=lfs diff=lfs merge=lfs -text
|
| 91 |
+
sc__healthy-nat__liver__13083cx33694g__GSM4648565.h5ad filter=lfs diff=lfs merge=lfs -text
|
| 92 |
+
visium__hcc-antiPD1__liu2025-1pts__2453cx36601g__GSE238264-HCC7NR.h5ad filter=lfs diff=lfs merge=lfs -text
|
| 93 |
+
sc__healthy__hepatocyte-v2__13635cx32596g__6b241dde-25c0-4edc-9021-03603ec3a524.h5ad filter=lfs diff=lfs merge=lfs -text
|
| 94 |
+
sc__hcc-mash-spectrum__huang2025-4pts__34396cx32285g__GSE282630.h5ad filter=lfs diff=lfs merge=lfs -text
|
| 95 |
+
sc__hcc-tcell__zheng2017-6pts__5063cx23459g__GSE98638.h5ad filter=lfs diff=lfs merge=lfs -text
|
| 96 |
+
sn__hcc-tumor-normal__alvarez2022-3pts__39995cx58100g__GSE189175.h5ad filter=lfs diff=lfs merge=lfs -text
|
| 97 |
+
visium__healthy-C73__blockC1__4992sx35477g__b2287ef1-eac3-49cc-93de-65df74e26a61.h5ad filter=lfs diff=lfs merge=lfs -text
|
| 98 |
+
visium__hcc-antiPD1__liu2025-1pts__3006cx36601g__GSE238264-HCC1R.h5ad filter=lfs diff=lfs merge=lfs -text
|
| 99 |
+
sn__hcc-tumor-normal__alvarez2022-1pts__39995cx58100g__GSE189175.h5ad filter=lfs diff=lfs merge=lfs -text
|
AI_powered_CCRM_liver.pdf
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e4aa301d634adf61a0af99689c444783c18a0994020faac05fe25ce07f034b90
|
| 3 |
+
size 10512624
|
AI_powered_CCRM_liver__reading_notes.md
ADDED
|
@@ -0,0 +1,180 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# 阅读笔记 · AI_powered_CCRM_liver.pdf
|
| 2 |
+
|
| 3 |
+
**生成日期**：2026-05-15
|
| 4 |
+
**对象文件**：本目录下的 `AI_powered_CCRM_liver.pdf`（10.5 MB / 21 页）
|
| 5 |
+
**目的**：作为 Liver_References 数据集的"理论锚点"——这篇论文不直接消费这 17 份数据，但它提出的框架决定了这批数据"为什么有用、未来怎么用"。
|
| 6 |
+
|
| 7 |
+
---
|
| 8 |
+
|
| 9 |
+
## 一、论文身份与定位
|
| 10 |
+
|
| 11 |
+
**标题**：*AI-powered critical care regenerative hepatology: adaptive repair for liver cancer and liver diseases*
|
| 12 |
+
|
| 13 |
+
**作者**：Xingcai Zhang¹·² · Frank Fu¹ · Isa Guo¹ · Richi Chen¹ · Aseel Malallah¹ · Natalie Torok²
|
| 14 |
+
（¹UCSD 化工与纳米工程系 · ²Stanford 医学院）
|
| 15 |
+
|
| 16 |
+
**通讯作者**：Zhang (xiz292@ucsd.edu) · Torok (ntorok@stanford.edu)
|
| 17 |
+
|
| 18 |
+
**论文类型**：**Perspective（视角/综述型论文）**——这是关键定性。Perspective 的写作惯例是综合现有文献提出一个新的组织框架，不报告原始实验数据，不做新的统计分析。论文末尾 `Data and code availability` 段明确写："No new datasets or custom code were generated in this Perspective." 这一句话直接决定了**论文与你 Liver_References 数据的关系**：数据不是论文的输入，而是论文框架未来操作化时的潜在底座。
|
| 19 |
+
|
| 20 |
+
**所属上位论文**：本文是 **"AI-CCRM"（AI-powered Critical Care Regenerative Medicine）系列的肝脏专属扩展**。母系列由 Zhang 等 2026 年在 *Science* 发表（参考文献 #1），把"AI 驱动的再生重症医学"作为通用框架。本文则把这个框架专门套到肝脏疾病上——这是一个**单器官化的延伸论文**。
|
| 21 |
+
|
| 22 |
+
---
|
| 23 |
+
|
| 24 |
+
## 二、核心主张
|
| 25 |
+
|
| 26 |
+
论文的中心论点用一句话讲：**严重肝病不应被当作静态诊断标签（MASH/ALD/ACLF/HCC/CCA），而应作为"再生 vs 错配修复 vs 免疫紊乱 vs 恶性可塑性"四种力量的动态竞争来处理；AI 是把多模态数据整合成一个状态判断引擎的工具**。
|
| 27 |
+
|
| 28 |
+
这个论点的几个支撑层次：
|
| 29 |
+
|
| 30 |
+
第一，**肝脏的悖论**——肝脏是哺乳动物里再生能力最强的器官（部分肝切除后剩余肝细胞能快速增殖恢复体积），但同样的可塑性也支持纤维化、肝硬化、HCC、CCA、肿瘤复发。**同一组生物学机制既是治愈的可能性，也是恶化的根源**——治疗设计的难点就在这里。
|
| 31 |
+
|
| 32 |
+
第二，**当前医疗的碎片化**——慢性肝病在门诊管理、失代偿在 ER/ICU 管理、肝癌在肿瘤科治疗、终末期等器官移植，"四套医生四套语言"，但**生物学层面这些状态是同一连续体**。论文主张 AI 是把这四套打通的胶水。
|
| 33 |
+
|
| 34 |
+
第三，**新的临床决策语法**——不再问"这个患者是 MASH 还是 ACLF 还是 HCC"，而是问五个问题：(a) 现在活跃的修复状态是哪种？(b) 这个状态可逆吗？(c) 它倾向于纤维化还是癌变？(d) 什么干预能把它推回去？(e) 治疗过程中状态变了，要怎么 adapt？
|
| 35 |
+
|
| 36 |
+
---
|
| 37 |
+
|
| 38 |
+
## 三、最有"齿"的设计：八个 AI-defined 肝修复状态
|
| 39 |
+
|
| 40 |
+
整篇 Perspective 最具体、最可操作的部分是**这八个修复状态的本体（ontology）**。它不是诊断、不是疾病分类，而是一组**正交于诊断标签的状态向量**，理论上一个患者可以同时处于多个状态。
|
| 41 |
+
|
| 42 |
+
| # | 修复状态 | 关键信号 | 典型临床情境 | 候选 AI 引导干预 |
|
| 43 |
+
|---|---|---|---|---|
|
| 44 |
+
| 1 | 急性炎症损伤 | DAMPs · 细胞因子 · ALT/AST 升高 · 线粒体应激 | ALF · 酒精性肝炎 · ACLF · 脓毒症肝损伤 | 炎症控制 · 线粒体保护 · 人工肝支持 · 决定再生治疗时机 |
|
| 45 |
+
| 2 | 再生允许损伤 | IL-6–STAT3 · HGF · EGF · 肝细胞增殖标志 | 肝切除后 · 早期 ALF · 早期 ACLF | MSC/EV · mRNA 营养因子 |
|
| 46 |
+
| 3 | 内皮 capillarization 主导损伤 | LSEC 去窗孔 · 内皮去分化 | 早中期纤维化 · 门脉高压前期 | LSEC-靶向 miRNA 球（参考文献 #8 Liu 2025） |
|
| 47 |
+
| 4 | 纤维化主导 maladaptive 修复 | 肝星状细胞激活 · 基质沉积 · 门脉高压 | 进展期纤维化 · 代偿期肝硬化 | 抗纤维化组合疗法 · senolytic |
|
| 48 |
+
| 5 | 免疫麻痹 / 耗竭 | CD8 衰竭 · IL-10 主导 · 感染易感 | ACLF · 失代偿期肝硬化 | 免疫调节 · 感染控制 · MSC/EV |
|
| 49 |
+
| 6 | 肿瘤–再生冲突 | 再生信号同时支持 HCC/CCA 生长 | HCC 在肝硬化背景 · 切除后复发 | "Cancer-safe" 再生策略 · 选择性 ICI |
|
| 50 |
+
| 7 | 移植桥接状态 | 多器官衰竭 · native 恢复不确定 | ACLF grade 2/3 · 严重 ALF | 人工肝支持 · 优先移植 |
|
| 51 |
+
| 8 | 治疗后重建 | 残余肝功能 · 复发风险评估 | 切除后 · 消融后 · 移植后 · ICU 出院后 | 监测 · 抗复发 · 抗纤维化预防 |
|
| 52 |
+
|
| 53 |
+
这套本体的好处是**可被算法学习**——给一个患者多模态数据，模型可以输出"在八个状态上各占多少比例"的概率向量，而不是非黑即白的诊断。但它的**风险**是过于抽象、临床医生可能拒绝接受；论文也承认这个 ontology 是"提议"（proposed），尚未经过同行验证或前瞻性使用。
|
| 54 |
+
|
| 55 |
+
---
|
| 56 |
+
|
| 57 |
+
## 四、技术栈五层（论文给出的工程蓝图）
|
| 58 |
+
|
| 59 |
+
论文把 AI 驱动的再生肝病学拆成五个工程层，每一层都标了代表性引用作"已有零件库"：
|
| 60 |
+
|
| 61 |
+
**第一层 · 生物材料**——空间修复架构。Nature-inspired 微图案、骨再生材料、黑磷凝胶、AI 辅助抗菌支架、生物来源材料。用于工程化肿瘤旁基质、抗纤维化深部、胆管修复支架、类器官培养系统、体外灌注装置（参考文献 #11–16）。
|
| 62 |
+
|
| 63 |
+
**第二层 · 治疗工程**——AI 驱动的纳米医学。**这一层提到 "bioorthogonal virus immuno-nanomedicine"——这就是 BOVIN（你 bovin-pathway-demo 项目）的指代**。同层还有热免疫纳米医学、巨噬细胞靶向促消解纳米药、跨血管纳米载体（参考文献 #9, #17–21）。
|
| 64 |
+
|
| 65 |
+
**第三层 · 计算与数字孪生**——基于现有 ICU 数据、肝病生物标志、影像、数字病理、组学、肠道菌群、EV cargo、设备衍生信号构建患者级"肝脏数字孪生"，模拟竞争性未来（native 恢复 vs 进展到 ACLF vs 转 HCC vs 抗纤维化响应 vs 切除后复发 vs 移植获益 / 无效）。
|
| 66 |
+
|
| 67 |
+
**第四层 · 传感**——多模态采集。包括 ICU 实时生理、可穿戴、可植入或可吞咽的生物电子、微流控床旁检测、AI 终端 chip。让数字孪生从"每周更新"变成"分钟级更新"，对 ACLF / 脓毒症肝损伤 / 移植后不稳定特别关键（参考文献 #27–34）。
|
| 68 |
+
|
| 69 |
+
**第五层 · 决策支持与闭环**——患者特异性类器官 + 多种治疗候选并行测试 + AI 量化形态/动力学/组学响应 + 选出"既能修复健康肝 + 又能抑制肿瘤 + 又能减少纤维化 + 又能保护免疫 + 又低毒"的方案。论文称之为 **"Cancer-safe liver regeneration testing architecture"**（Figure 3）。
|
| 70 |
+
|
| 71 |
+
---
|
| 72 |
+
|
| 73 |
+
## 五、Cancer-safe 再生——论文最有原创性的概念
|
| 74 |
+
|
| 75 |
+
这是整篇 Perspective **最具原创性、最值得记住**的概念。传统再生医学的核心命题是"促进再生"，几乎不考虑癌变风险。但**肝脏再生的分子通路（Wnt、Notch、TGF-β、HGF、IL-6/STAT3、肝星状细胞激活）与 HCC 致癌通路高度重叠**——盲目促再生很可能同时促癌。
|
| 76 |
+
|
| 77 |
+
论文提议的解法是 **"配对类器官平台"**：
|
| 78 |
+
|
| 79 |
+
> 同一患者来源 → 提取四种东西平行培养：
|
| 80 |
+
>
|
| 81 |
+
> (a) 非恶性肝实质类器官（正常肝细胞 + 胆管细胞）
|
| 82 |
+
> (b) 肿瘤类器官（如果患者有 HCC/CCA）
|
| 83 |
+
> (c) 免疫细胞 + 肝星状细胞 + 内皮细胞共培养
|
| 84 |
+
> (d) 患者来源胞外基质重建
|
| 85 |
+
>
|
| 86 |
+
> 把候选治疗丢到这四个系统上同时跑，AI 评估**五个输出**：
|
| 87 |
+
>
|
| 88 |
+
> 1. 健康肝组织修复程度
|
| 89 |
+
> 2. 肿瘤生长抑制程度
|
| 90 |
+
> 3. 纤维化程度
|
| 91 |
+
> 4. 免疫功能保留/恢复程度
|
| 92 |
+
> 5. 系统毒性
|
| 93 |
+
|
| 94 |
+
**只有在五个维度都通过的治疗才进入下一步临床**——这是"cancer-safe"的工程化定义。论文承认这套系统在 2026 年还是 aspirational（远景目标），但提出了具体的实施清单。
|
| 95 |
+
|
| 96 |
+
这个概念对 BOVIN 项目的直接含义：BOVIN 的 11 模块（M1–M11）输出的不应该只是"ICI 响应预测"，未来还应该输出"修复 vs 致癌"的双向风险评估——这是 BOVIN Aim 3（从预测到处方）的潜在主轴。
|
| 97 |
+
|
| 98 |
+
---
|
| 99 |
+
|
| 100 |
+
## 六、与你 Liver_References 17 份数据的对应关系
|
| 101 |
+
|
| 102 |
+
这是这份阅读笔记最核心的章节——论文虽然不直接用这 17 份数据，但**论文的框架决定了为什么收集这批数据**。
|
| 103 |
+
|
| 104 |
+
| 论文中的需求 | Liver_References 提供了什么 | 缺什么 |
|
| 105 |
+
|---|---|---|
|
| 106 |
+
| 修复状态 3（LSEC capillarization 主导） | `sc__healthy__endothelial__9422cx32596g__a95e1659...h5ad` 含 9,422 个健康肝内皮细胞，是定义"健康 LSEC 表型"的基线 | 需要纤维化/肝硬化患者的 LSEC 数据，目前只有 PSC/PBC 的 sn 全细胞图谱包含少量 |
|
| 107 |
+
| 修复状态 4（纤维化主导） | `sc__healthy__stellate__1417cx32596g__08a9f031...h5ad` 含 1,417 个健康肝星状细胞，是"静默状态"基线 | 需要"激活态肝星状细胞"数据——本 collection 无独立切片 |
|
| 108 |
+
| 修复状态 1+2（炎症/再生允许） | `sc__healthy__hepatocyte-v1__53015cx32596g__63137dcf...h5ad` 5.3 万健康肝细胞（含 zonation 三亚型） | 需要损伤/再生状态肝细胞——本 collection 无 |
|
| 109 |
+
| 修复状态 6（肿瘤-再生冲突） | **无** —— 本 collection 不含 HCC 肿瘤数据 | 需要 HCC 肿瘤 scRNA（如 Ma 2021 GSE149614），登记在 bovin-bench/hcc/CATALOG.md FOUND-05 |
|
| 110 |
+
| 论文提到的胆管细胞修复 | `sc__healthy__cholangiocyte__1011cx32596g__601ef580...h5ad` 含 1,011 健康胆管细胞 | PSC 患者的病变胆管细胞数据散落在 sn 全细胞图谱中 |
|
| 111 |
+
| PSC 病变直接对照 | `sn__psc-pbc-healthy__all-cells__105780cx32596g__4b5895d7...h5ad` 含 47,156 PSC 细胞 vs 26,515 健康，可直接做病例-对照 | 缺 sc（cell）版本的 PSC 配对（只有 sn） |
|
| 112 |
+
| 论文提到��空间修复（biomaterial 层） | 7 份 Visium 文件提供空间锚定——3 切片 C73 健康 + 4 切片 PSC011 | Visium 仅来自 2 个供体，统计自由度极低；C73 与 PSC011 不是同一组人，跨患者对比有限 |
|
| 113 |
+
|
| 114 |
+
**核心观察**：你这 17 份数据**正好覆盖了论文"健康基线"的所有 8 种关键细胞类型 + 提供了一个"慢性胆汁淤积病"对照（PSC/PBC）**。从论文框架角度看，这批数据是 **"修复状态 3 (内皮) + 修复状态 4 (纤维化) 的 baseline 端 + 修复状态 5 (免疫) 的健康端"** 的最佳现成资源。要把论文框架全部跑通，还需要补的是：
|
| 115 |
+
|
| 116 |
+
> 一，**HCC / CCA 肿瘤 scRNA 数据**——用于建模修复状态 6（肿瘤-再生冲突）。bovin-bench/hcc/CATALOG.md 已经列出（FOUND-05 / FOUND-06），但数据尚未下载到 database_unified/。
|
| 117 |
+
>
|
| 118 |
+
> 二，**急性损伤期肝脏数据**（ACLF / ALF / 酒精性肝炎）——用于建模修复状态 1 + 2。这类数据较稀缺，多在 EGA / dbGaP 受控访问。
|
| 119 |
+
>
|
| 120 |
+
> 三，**激活态肝星状细胞 / capillarized LSEC 数据**——用于建模修复状态 3 + 4 的"病态端"。可能要从其他 scRNA paper 单独采集。
|
| 121 |
+
>
|
| 122 |
+
> 四，**纵向 / 系列样本**——患者治疗前后配对。是论文中提到的"修复状态向量随时间演变"的必要条件，目前 collection 全是横截面。
|
| 123 |
+
|
| 124 |
+
---
|
| 125 |
+
|
| 126 |
+
## 七、与 BOVIN 项目的关系
|
| 127 |
+
|
| 128 |
+
BOVIN（**B**ioorthogonal **O**ptimized **V**irus **I**mmuno-**N**anomedicine）在这篇 Perspective 里**被点名提到**——参考文献 #19 / #21 区段，"bioorthogonal virus immuno-nanomedicine" 作为"治疗工程层"的代表性技术之一。论文的语气是**把 BOVIN 当作一个具体技术示例**放进更宏大的肝病学 AI-CCRM 框架。
|
| 129 |
+
|
| 130 |
+
这意味着 BOVIN 项目（bovin-pathway-demo + bovin-bench）在 Zhang & Torok 的整体科研路线图里是**"治疗工程层"的一个具体落点**。Liver_References 数据则是**"传感 + 计算层"的一个数据底座**。两者在论文框架内是配套的——治疗工程提供"会发生什么"的工具，传感+计算层提供"对谁会发生什么"的判断。
|
| 131 |
+
|
| 132 |
+
具体落到你目前正在做的事：
|
| 133 |
+
|
| 134 |
+
> **bovin-pathway-demo 的 11 模块** ↔ 可对应到论文中的"修复状态向量"——M6 (IFN) 对应炎症/免疫激活，M4 (DAMP) 对应损伤释放，M7 (antigen presentation) 对应免疫识别。BOVIN 现有的 IG 归因可以被解读为"这个患者目前的修复状态分布"。
|
| 135 |
+
>
|
| 136 |
+
> **bovin-bench 的 ICI 队列** ↔ 论文中"修复状态 6（肿瘤-再生冲突）"的实证数据集——既有肿瘤（需要免疫激活）也有 cirrhotic background（需要免疫克制 + 抗纤维化）。
|
| 137 |
+
>
|
| 138 |
+
> **未来 Aim 3（OV 设计回路）** ↔ 论文中"Cancer-safe 再生测试架构"的实例化——4 种类器官 × 5 输出的工程实现。
|
| 139 |
+
|
| 140 |
+
所以这篇 Perspective 的真正价值，是它**把 BOVIN 从一个孤立的纳米药物项目，重新框架成 AI-CCRM 整体生态里的一个具体组件**。这对论文写作（说服评审 BOVIN 不只是 ad hoc 工程）、对 grant 申请（讲清楚 BOVIN 在哪个更大的 vision 里）、对 collaborator pitch（Zhang + Torok 的双 lab 合作）都很有用。
|
| 141 |
+
|
| 142 |
+
---
|
| 143 |
+
|
| 144 |
+
## 八、值得注意的几点（含批判性观察）
|
| 145 |
+
|
| 146 |
+
第一，**"No new datasets" 的双刃剑**——论文坦诚自己是 Perspective，但这也意味着**框架未被任何具体数据验证**。8 个修复状态是"提议"，5 阶段 roadmap 是"愿景"，cancer-safe 标准是"应当"。读者要警惕"听起来合理 ≠ 经过验证"。建议在自己之后的论文里**给框架某个具体部分做出可测试的实证**——比如用 BOVIN 数据示范"在 ICI 队列里，BOVIN 的 11 模块输出可以被映射到论文 8 个修复状态的概率分布上"——这就把 Perspective 的 conceptual 主张落到 empirical 层面。
|
| 147 |
+
|
| 148 |
+
第二，**自引网络浓度较高**——参考文献 #1（AI-CCRM 母 Science 2026）、#9（AI-nanomedicine Chem Soc Rev 2026）、#23（mitochondrial phenotype Nat Comput Sci 2024）、#28（smartphone microfluidic Nat Commun 2023）等都是 Zhang 自己的论文。**这本身不是问题**——他确实在这个方向做了很多事——但**外部 paper 链接的密度可以更高**。如果是论文评审，可能会评一句 "the literature foundation could be broader"。
|
| 149 |
+
|
| 150 |
+
第三，**8 个修复状态的粒度选择**——8 个状态比"4 类疾病诊断"细，比"30 个分子亚型"粗。这个粒度是论文作者的工程取舍，不是从数据反推出来的。**未来真实数据可能支持的状态数 ≠ 8**——可能是 5 个、可能是 12 个。表 1 里这 8 个的"代表性信号"列也偏概念性、不够定量。建议你后续做实证工作时，**不要锁死在 8 这个数字上**，让数据告诉你真实的 state 数。
|
| 151 |
+
|
| 152 |
+
第四，**临床医生接受度风险**——把诊断标签替换成 8 维状态向量，这在 AI/ML 领域是自然的，但在临床端是巨大的认知转换。论文承认这一点（"AI models must be explainable, bias-tested and integrated into clinician workflows without replacing clinical judgment"），但没给出具体的过渡方案。**这是 Perspective 转化为产品时最大的非技术阻力**。
|
| 153 |
+
|
| 154 |
+
第五，**5 阶段 roadmap 的"Stage 1"是关键卡点**——Stage 1（"Retrospective integration of large clinical, imaging, histology, omics and outcome datasets"）写一句话，但实际工作量是巨大的。Liver_References 17 份只是这个 Stage 1 的一小部分；要把 MASLD / ALD / ACLF / HCC / CCA / 移植后等所有这些场景的数据都拉齐，**单是数据收集 + 清洗就是 3–5 年工程量**。论文没有给出 Stage 1 的人力 / 资金估计——这往往是 grant 评审会问到的事。
|
| 155 |
+
|
| 156 |
+
第六，**"Bioorthogonal virus immuno-nanomedicine" 在论文里的出场是 reference-level 的**——意思是 BOVIN 是论文引用的众多技术之一，不是论文的主体。这对你的意义是：**这篇论文是 Zhang 给 BOVIN 写的"广告位"，不是 BOVIN 自己的成果论文**。BOVIN 自己的 first-authorship paper（你 Frank Fu 作为可能的第一作者）还需要另写——大概率会用 bovin-pathway-demo 的 v0.2-aim2 结果 + 一份新的 cancer-safe 验证（用 Liver_References 数据 + 未来 HCC tumor 数据配对类器官）。
|
| 157 |
+
|
| 158 |
+
---
|
| 159 |
+
|
| 160 |
+
## 九、对你"下一步"的几个具体启示
|
| 161 |
+
|
| 162 |
+
第一，**Liver_References 数据集的"使用价值定位"应该明确**——它不是 ICI 响应预测的训练集（那是 bovin-bench 的工作），它是论文框架里的 **"修复状态 baseline reference"**。对它的使用应该围绕"什么是健康肝 + 慢性胆汁淤积病的细胞状态"这个问题展开，而不是"它怎么帮 BOVIN 提升 AUC"。两条路径都对，但思路不同。
|
| 163 |
+
|
| 164 |
+
第二，**优先补齐 HCC 肿瘤数据**——论文最有原创性的部分（cancer-safe 再生）需要肿瘤 + 非肿瘤的配对。Liver_References 提供了非肿瘤一半，bovin-bench/hcc/CATALOG.md 的 FOUND-05 / FOUND-06 提供了肿瘤一半的"已登记但未下载"清单。把这些数据拉到 `database_unified/HCC_Tumor_Cohorts/` 是非常自然的下一步——可以复用 Liver_References 的命名规则（modality__cohort__content__nxm__uuid.h5ad）。
|
| 165 |
+
|
| 166 |
+
第三，**考虑做一份"框架可测试性"小论文**——把这篇 Perspective 的某个具体主张拿出来用实证数据测一遍。最容易的标的：把 BOVIN 在 v0.2-aim2 的 IG 归因映射到论文 8 个修复状态上，看看 ICI 响应者与非响应者在状态空间里的分布是否可分。这种"小落地"论文是 Zhang lab 经典做法的延伸（先 Perspective 立框架、后 empirical paper 实例化）。
|
| 167 |
+
|
| 168 |
+
第四，**作为 AIVIN 的 README**——这篇 Perspective 实际上是整个 AIVIN 项目（bovin-pathway-demo + bovin-bench + database_unified）的**最高层 README**。它告诉任何读者"为什么这三个子项目存在、它们如何拼成一个整体"。建议在 AIVIN 顶层的某个文档（比如未来可能的 `AIVIN/README.md`）里把这篇 Perspective 的 8 状态 + 5 阶段 + cancer-safe 三个核心概念 link 进来，作为项目"为什么"的来源。
|
| 169 |
+
|
| 170 |
+
第五，**Zhang + Torok 的双 lab 合作意义**——论文第一作者 Zhang 是 UCSD（你的导师），通讯作者 Torok 是 Stanford 医学院。这意味着 **AIVIN 项目背后是双校合作**，未来如果要去 Stanford 拿 sample / clinical data，Torok lab 是直接通道。这对 bovin-bench 里那些标注"DUA / sponsor-only"的 HCC ICI 队列尤其重要——Stanford 医学院通常能走 academic DUA 路径访问 IMbrave150 / KEYNOTE-224 等数据集。
|
| 171 |
+
|
| 172 |
+
---
|
| 173 |
+
|
| 174 |
+
## 十、一句话总结
|
| 175 |
+
|
| 176 |
+
> 这篇 Perspective 是 AIVIN 项目的**理论母舰**——它不直接用 Liver_References 数据，但它提出的 8 个修复状态 + 5 层技术栈 + cancer-safe 范式，**决定了你应该用什么数据 + 怎么用 + 用来回答什么问题**。BOVIN 在论文里只是众多"治疗工程层"技术之一；Liver_References 是论文"修复状态 baseline"层的第一块拼图；未来 HCC_Tumor_Cohorts、organoid 数据、ACLF 患者数据等都将作为这个框架的其他拼图陆续加入。
|
| 177 |
+
|
| 178 |
+
---
|
| 179 |
+
|
| 180 |
+
*阅读笔记生成自 PDF 21 页全文的逐节解读 · 关联到 Liver_References 17 份数据 · 关联到 bovin-pathway-demo 与 bovin-bench 现有进度 · 笔记本身不修改源 PDF，只作伴随文档。*
|
README.md
ADDED
|
@@ -0,0 +1,281 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: cc-by-4.0
|
| 3 |
+
language:
|
| 4 |
+
- en
|
| 5 |
+
size_categories:
|
| 6 |
+
- 1M<n<10M
|
| 7 |
+
task_categories:
|
| 8 |
+
- feature-extraction
|
| 9 |
+
tags:
|
| 10 |
+
- single-cell
|
| 11 |
+
- scRNA-seq
|
| 12 |
+
- snRNA-seq
|
| 13 |
+
- spatial-transcriptomics
|
| 14 |
+
- visium
|
| 15 |
+
- cancer
|
| 16 |
+
- hepatocellular-carcinoma
|
| 17 |
+
- HCC
|
| 18 |
+
- iCCA
|
| 19 |
+
- liver
|
| 20 |
+
- atlas
|
| 21 |
+
- cellxgene
|
| 22 |
+
- anndata
|
| 23 |
+
pretty_name: AIVIN Liver References
|
| 24 |
+
---
|
| 25 |
+
|
| 26 |
+
# AIVIN · Liver References
|
| 27 |
+
|
| 28 |
+
|
| 29 |
+
|
| 30 |
+
Harmonized single-cell + single-nucleus + Visium spatial reference set for
|
| 31 |
+
**human liver tissue** (cancer + healthy + chronic-liver-disease), curated
|
| 32 |
+
under the AIVIN cross-cancer reference project at UCSD. All files are
|
| 33 |
+
[CELLxGENE schema 7.0.0](https://github.com/chanzuckerberg/single-cell-curation)
|
| 34 |
+
compatible and follow the unified AIVIN naming + provenance convention.
|
| 35 |
+
|
| 36 |
+
| | |
|
| 37 |
+
|---|---|
|
| 38 |
+
| **Files** | **41** `.h5ad` (40 valid + 1 known-broken — see Known Issues) |
|
| 39 |
+
| **Cells (sc + sn)** | **~1,177,000** |
|
| 40 |
+
| **Visium spots** | **~34,900** |
|
| 41 |
+
| **Distinct cohorts** | **18** (12 cancer + 5 healthy + 1 CLD) |
|
| 42 |
+
| **Source studies** | 15 GEO + CELLxGENE Census + 1 GSM-direct |
|
| 43 |
+
| **Disease dimensions** | HCC primary · HCC metastatic · HCC fetal · HCC anti-PD1 · HCC MASH · iCCA · cholangiocarcinoma · NASH · HCV · chronic liver disease · PSC · PBC · healthy |
|
| 44 |
+
| **Platforms** | 10x Chromium v2/v3 · Smart-seq2 · CEL-Seq2 · 10x Visium |
|
| 45 |
+
| **Total size** | ~14 GB |
|
| 46 |
+
| **Snapshot** | AIVIN 2026-Q2 v1.0 |
|
| 47 |
+
| **Zenodo DOI** | `[pending — Sat 5/30 snapshot]` |
|
| 48 |
+
| **HF DOI** | `[pending — mint after upload]` |
|
| 49 |
+
| **License** | CC-BY-4.0 (plus cite original cohort papers) |
|
| 50 |
+
|
| 51 |
+
---
|
| 52 |
+
|
| 53 |
+
## What's in this repo
|
| 54 |
+
|
| 55 |
+
Every `.h5ad` follows the AIVIN naming convention (see `NAMING_INDEX §八` in the AIVIN GitHub for the full grammar):
|
| 56 |
+
|
| 57 |
+
```
|
| 58 |
+
<modality>__<cohort-slug>__<who>__<NcxMg>__<accession>.h5ad
|
| 59 |
+
│ │ │ │ └── GEO GSE / GSM · GSA HRA · CELLxGENE UUID · Zenodo ID
|
| 60 |
+
│ │ │ └────────────── shape: cells × genes (or spots × genes for visium)
|
| 61 |
+
│ │ └────────────────────── first-author + year + n donors/samples
|
| 62 |
+
│ └─────────────────────────────────── biology tag (disease + sub-type)
|
| 63 |
+
└───────────────────────────────────────────── modality: sc = single-cell · sn = single-nucleus · visium = spatial
|
| 64 |
+
```
|
| 65 |
+
|
| 66 |
+
### Cohort manifest
|
| 67 |
+
|
| 68 |
+
#### Cancer cohorts (GEO source · 24 files · ~874k cells)
|
| 69 |
+
|
| 70 |
+
| Cohort slug | Source | Cells | Disease | Platform | Citation |
|
| 71 |
+
|---|---|---:|---|---|---|
|
| 72 |
+
| `hcc-cd45` | GSE235863 | 191,435 | HCC CD45+ enriched | 10x | Guo et al., 2025 |
|
| 73 |
+
| `hcc-fetal` | GSE156625 | 109,238 | HCC onco-fetal | 10x | Sharma et al., *Cell* 2020 |
|
| 74 |
+
| `hcc-cd8tcell` | GSE235863 | 95,408 | HCC CD8 T cells | 10x | Guo et al., 2025 |
|
| 75 |
+
| `hcc-tumor-normal` (Sharma) | GSE156625 | 73,589 | HCC tumor + adjacent | 10x | Sharma et al., *Cell* 2020 |
|
| 76 |
+
| `hcc-multisite` | GSE149614 | 71,915 | HCC primary + metastatic + PVTT + LN | 10x | Lu et al., *Nat Commun* 2022 |
|
| 77 |
+
| `hcc-iccA-cd45` | GSE140228-droplet | 66,187 | HCC + iCCA, CD45+ | 10x | Sharma et al., *Cell* 2020 |
|
| 78 |
+
| `hcc-iccA-treated` | GSE151530 | 56,721 | HCC + iCCA, post-treatment | 10x | Ma et al., *J Hepatol* 2021 |
|
| 79 |
+
| `hcc-trm` | GSE281110 | 41,848 | HCC tumor-associated TRM T | 10x | Park et al., 2025 |
|
| 80 |
+
| `hcc-tumor-normal-3pt` | GSE189175 | 39,995 | HCC tumor + normal | sn-10x | Alvarez et al., 2022 |
|
| 81 |
+
| `hcc-tumor-normal-1pt` | GSE189175 | 39,995 | (duplicate — see Known Issues) | sn-10x | Alvarez et al., 2022 |
|
| 82 |
+
| `hcc-mash-spectrum` | GSE282630 | 34,396 | HCC + MASH spectrum | 10x | Huang et al., 2025 |
|
| 83 |
+
| `hcc-cd45-ss2` | GSE140228-ss2 | 7,074 | HCC CD45+ Smart-seq2 | SS2 | Sharma et al., *Cell* 2020 |
|
| 84 |
+
| `hcc-iccA-mixed-set1` | GSE125449-set1 | 5,115 | HCC + iCCA, mixed | 10x | Ma et al., *Cancer Cell* 2019 |
|
| 85 |
+
| `hcc-tcell` | GSE98638 | 5,063 | HCC infiltrating T cells | SMART-seq2 | Zheng et al., *Cell* 2017 |
|
| 86 |
+
| `hcc-iccA-mixed-set2` | GSE125449-set2 | 4,831 | HCC + iCCA, mixed | 10x | Ma et al., *Cancer Cell* 2019 |
|
| 87 |
+
| `hcc-antiPD1` (R1) | GSE238264-HCC1R | 3,006 | HCC anti-PD1 responder | 10x | Liu et al., 2025 |
|
| 88 |
+
| `hcc-antiPD1` (R4) | GSE238264-HCC4R | 3,002 | HCC anti-PD1 responder | 10x | Liu et al., 2025 |
|
| 89 |
+
| `hcc-antiPD1` (R2) | GSE238264-HCC2R | 2,766 | HCC anti-PD1 responder | 10x | Liu et al., 2025 |
|
| 90 |
+
| `hcc-antiPD1` (NR6) | GSE238264-HCC6NR | 2,575 | HCC anti-PD1 non-responder | 10x | Liu et al., 2025 |
|
| 91 |
+
| `hcc-antiPD1` (NR7) | GSE238264-HCC7NR | 2,453 | HCC anti-PD1 non-responder | 10x | Liu et al., 2025 |
|
| 92 |
+
| `hcc-antiPD1` (R3) | GSE238264-HCC3R | 2,170 | HCC anti-PD1 responder | 10x | Liu et al., 2025 |
|
| 93 |
+
| `hcc-antiPD1` (NR5) | GSE238264-HCC5NR | 1,320 | HCC anti-PD1 non-responder | 10x | Liu et al., 2025 |
|
| 94 |
+
| `cld-lyec` | GSE129933 | 901 | Chronic liver disease lymphatic EC | SMART-seq2 | Tamburini et al., *Front Immunol* 2019 |
|
| 95 |
+
| `healthy-nat` | GSM4648565 | 13,083 | healthy liver | 10x | (Nat Commun 2020) |
|
| 96 |
+
|
| 97 |
+
#### Healthy + autoimmune baselines (CELLxGENE Census · 11 sc/sn files · ~303k cells)
|
| 98 |
+
|
| 99 |
+
| Cohort slug | Cells | Cell type / disease | Modality |
|
| 100 |
+
|---|---:|---|---|
|
| 101 |
+
| `psc-pbc-healthy` (sn) | 105,780 | PSC + PBC + healthy, all cells | sn |
|
| 102 |
+
| `psc-pbc-healthy` (sc) | 89,637 | PSC + PBC + healthy, all cells | sc |
|
| 103 |
+
| `healthy hepatocyte-v1` | 53,015 | hepatocytes | sc |
|
| 104 |
+
| `healthy lymphoid` | 16,665 | lymphoid lineage | sc |
|
| 105 |
+
| `healthy hepatocyte-v2` | 13,635 | hepatocytes (alt curation) | sc |
|
| 106 |
+
| `healthy macrophage` | 11,127 | macrophages | sc |
|
| 107 |
+
| `healthy endothelial` | 9,422 | endothelial cells | sc |
|
| 108 |
+
| `healthy stellate` | 1,417 | hepatic stellate cells | sc |
|
| 109 |
+
| `healthy b-cell` | 1,250 | B cells | sc |
|
| 110 |
+
| `healthy cholangiocyte` | 1,011 | cholangiocytes | sc |
|
| 111 |
+
|
| 112 |
+
#### Spatial transcriptomics (CELLxGENE Census · 6 Visium files · ~35k spots)
|
| 113 |
+
|
| 114 |
+
| Cohort slug | Spots | Tissue block | Disease |
|
| 115 |
+
|---|---:|---|---|
|
| 116 |
+
| `visium healthy-C73 / blockA1` | 4,992 | block A1 | healthy donor C73 |
|
| 117 |
+
| `visium healthy-C73 / blockC1` | 4,992 | block C1 | healthy donor C73 |
|
| 118 |
+
| `visium healthy-C73 / blockD1` | 4,992 | block D1 | healthy donor C73 |
|
| 119 |
+
| `visium psc-PSC011 / blockA1` | 4,992 | block A1 | PSC patient 011 |
|
| 120 |
+
| `visium psc-PSC011 / blockB1` | 4,992 | block B1 | PSC patient 011 |
|
| 121 |
+
| `visium psc-PSC011 / blockC1` | 4,992 | block C1 | PSC patient 011 |
|
| 122 |
+
| `visium psc-PSC011 / blockD1` | 4,992 | block D1 | PSC patient 011 |
|
| 123 |
+
|
| 124 |
+
---
|
| 125 |
+
|
| 126 |
+
## Schema
|
| 127 |
+
|
| 128 |
+
All `.h5ad` conform to **CELLxGENE schema 7.0.0** plus AIVIN extensions:
|
| 129 |
+
|
| 130 |
+
**`obs` (cells) — required columns**
|
| 131 |
+
|
| 132 |
+
- `cell_id` (index)
|
| 133 |
+
- `donor_id` (when known)
|
| 134 |
+
- `tissue_site` — unified vocab: `PT` (primary tumor) · `NTL` (normal liver) · `JTL` (juxta-tumor liver) · `MLN` (lymph node metastasis) · `PVTT` (portal vein tumor thrombus) · `PBMC` (peripheral blood) · `LIL` (liver intra-lesional)
|
| 135 |
+
- `disease` — values within the **Disease dimensions** list above
|
| 136 |
+
- `cell_type` (when annotated by original author)
|
| 137 |
+
- `assay` — platform / chemistry
|
| 138 |
+
|
| 139 |
+
**`var` (genes) — convention**
|
| 140 |
+
|
| 141 |
+
- Ensembl ID as `var.index` (when available, esp. CELLxGENE-sourced)
|
| 142 |
+
- Some GEO-sourced cohorts use HGNC `gene_symbol` as index + `entrez_id` column
|
| 143 |
+
- Heterogeneity across cohorts: 18 distinct gene-space sizes (2,384 – 58,100 genes) — see `aivin_obs_field_notes` per file for caveats; downstream concat use `ad.concat(..., join='outer')`
|
| 144 |
+
|
| 145 |
+
**`uns` (provenance, AIVIN-specific)**
|
| 146 |
+
|
| 147 |
+
- `citation` — full APA reference
|
| 148 |
+
- `doi` — primary paper DOI
|
| 149 |
+
- `source_accession` — GEO GSE / GSM / GSA HRA / CELLxGENE UUID / Zenodo ID
|
| 150 |
+
- `source_url`
|
| 151 |
+
- `aivin_ingest_date`
|
| 152 |
+
- `aivin_cohort_slug`
|
| 153 |
+
- `aivin_source_files` — original raw filename list
|
| 154 |
+
- `aivin_obs_field_notes` — any value-mapping done in ingest
|
| 155 |
+
|
| 156 |
+
---
|
| 157 |
+
|
| 158 |
+
## Usage
|
| 159 |
+
|
| 160 |
+
### Load one cohort (lazy / single file)
|
| 161 |
+
|
| 162 |
+
```python
|
| 163 |
+
from huggingface_hub import hf_hub_download
|
| 164 |
+
import anndata as ad
|
| 165 |
+
|
| 166 |
+
path = hf_hub_download(
|
| 167 |
+
repo_id='AIVIN-UCSD/liver-references',
|
| 168 |
+
filename='sc__hcc-multisite__lu2022-10pts__71915cx25712g__GSE149614.h5ad',
|
| 169 |
+
repo_type='dataset',
|
| 170 |
+
)
|
| 171 |
+
a = ad.read_h5ad(path)
|
| 172 |
+
print(a)
|
| 173 |
+
# Inspect AIVIN provenance
|
| 174 |
+
print(a.uns['citation'])
|
| 175 |
+
print(a.uns['aivin_obs_field_notes'])
|
| 176 |
+
```
|
| 177 |
+
|
| 178 |
+
### Load all cancer cohorts + concat (gene union)
|
| 179 |
+
|
| 180 |
+
```python
|
| 181 |
+
from huggingface_hub import snapshot_download
|
| 182 |
+
from pathlib import Path
|
| 183 |
+
import anndata as ad
|
| 184 |
+
|
| 185 |
+
local = snapshot_download(
|
| 186 |
+
repo_id='AIVIN-UCSD/liver-references',
|
| 187 |
+
repo_type='dataset',
|
| 188 |
+
allow_patterns='sc__hcc-*.h5ad', # cancer only
|
| 189 |
+
ignore_patterns='*macparland2019-0donors*', # skip known-broken file
|
| 190 |
+
)
|
| 191 |
+
adatas = {f.stem: ad.read_h5ad(f) for f in Path(local).glob('sc__hcc-*.h5ad')}
|
| 192 |
+
merged = ad.concat(adatas, axis=0, join='outer', label='cohort', fill_value=0)
|
| 193 |
+
print(merged)
|
| 194 |
+
# ~750k cells × union of genes across cohorts
|
| 195 |
+
```
|
| 196 |
+
|
| 197 |
+
### Pipe into scvi-tools (foundation model training)
|
| 198 |
+
|
| 199 |
+
```python
|
| 200 |
+
import scvi
|
| 201 |
+
scvi.model.SCVI.setup_anndata(merged, batch_key='cohort')
|
| 202 |
+
model = scvi.model.SCVI(merged, n_layers=2, n_latent=30)
|
| 203 |
+
model.train(accelerator='mps') # Apple Silicon MPS acceleration
|
| 204 |
+
```
|
| 205 |
+
|
| 206 |
+
---
|
| 207 |
+
|
| 208 |
+
## Citation
|
| 209 |
+
|
| 210 |
+
If you use this dataset in a publication, please cite:
|
| 211 |
+
|
| 212 |
+
1. **AIVIN as a collection** (this dataset card):
|
| 213 |
+
```bibtex
|
| 214 |
+
@dataset{aivin_liver_2026Q2,
|
| 215 |
+
author = {AIVIN Project, UCSD},
|
| 216 |
+
title = {{AIVIN Liver References (2026-Q2 v1.0)}},
|
| 217 |
+
year = {2026},
|
| 218 |
+
publisher = {Hugging Face},
|
| 219 |
+
doi = {[pending HF DOI mint]},
|
| 220 |
+
url = {https://huggingface.co/datasets/AIVIN-UCSD/liver-references}
|
| 221 |
+
}
|
| 222 |
+
```
|
| 223 |
+
|
| 224 |
+
2. **Each individual cohort** — see the `uns.citation` field of every `.h5ad`,
|
| 225 |
+
or the **Cohort manifest** table above. Particularly for landmark papers:
|
| 226 |
+
- Lu et al., *Nat Commun* 13:4594 (2022) — `doi:10.1038/s41467-022-32283-3`
|
| 227 |
+
- Sharma et al., *Cell* 183:377 (2020) — `doi:10.1016/j.cell.2020.08.040`
|
| 228 |
+
- Ma et al., *J Hepatol* 75:1418 (2021) — `doi:10.1016/j.jhep.2021.06.028`
|
| 229 |
+
- Ma et al., *Cancer Cell* 36:418 (2019) — `doi:10.1016/j.ccell.2019.08.007`
|
| 230 |
+
- Zheng et al., *Cell* 169:1342 (2017) — `doi:10.1016/j.cell.2017.05.035`
|
| 231 |
+
|
| 232 |
+
3. **(Optional) the Zenodo permanent snapshot** for byte-frozen reproducibility:
|
| 233 |
+
`doi: [pending Sat 5/30]`
|
| 234 |
+
|
| 235 |
+
---
|
| 236 |
+
|
| 237 |
+
## License
|
| 238 |
+
|
| 239 |
+
This collection is released under **CC-BY-4.0**. The license applies to AIVIN's
|
| 240 |
+
harmonization, schema mapping, and provenance metadata. **You must still cite
|
| 241 |
+
the original cohort papers** when using their data — see the per-cohort
|
| 242 |
+
manifest above. Cohorts derived from controlled-access sources (e.g., GSA-Human
|
| 243 |
+
HRA001748 Xue 2022) are NOT included in this public repo; see the cross-tissue
|
| 244 |
+
meta-repo for access pointers.
|
| 245 |
+
|
| 246 |
+
---
|
| 247 |
+
|
| 248 |
+
## Pipeline & reproducibility
|
| 249 |
+
|
| 250 |
+
- **Ingest scripts**: `github.com/AIVIN-UCSD/aivin/tree/main/scripts`
|
| 251 |
+
(per-cohort `<Cn>_<author><year>_ingest.py` + `W3_backlog_ingest.py` dispatcher)
|
| 252 |
+
- **Methods extracts**: per-paper structured methods at
|
| 253 |
+
`github.com/AIVIN-UCSD/aivin/tree/main/literature/A_cancer_TME/methods_extracts`
|
| 254 |
+
- **Structure report**: full per-file schema audit at
|
| 255 |
+
`github.com/AIVIN-UCSD/aivin/blob/main/database_unified/Liver_References/STRUCTURE_REPORT.md`
|
| 256 |
+
- **Backlog inventory**: candidates for v3 (3-month) expansion at
|
| 257 |
+
`github.com/AIVIN-UCSD/aivin/blob/main/database_unified/_staging/BACKLOG_INVENTORY.md`
|
| 258 |
+
|
| 259 |
+
---
|
| 260 |
+
|
| 261 |
+
## Known issues (v1.0)
|
| 262 |
+
|
| 263 |
+
| Issue | Affected file | Fix planned |
|
| 264 |
+
|---|---|---|
|
| 265 |
+
| **MacParland v1 ingest broken** (shape `0 × 3,958,008`) — the multi-plate CEL-Seq2 concat in `ingest_GSE124395()` produced a degenerate output | `sc__healthy-hlca__macparland2019-0donors__0cx3958008g__GSE124395.h5ad` | Will re-ingest in v1.1 with proper plate-level dedup; **filter out via `ignore_patterns='*macparland2019-0donors*'`** in `snapshot_download` |
|
| 266 |
+
| **GSE189175 Alvarez duplicate** — same 39,995 cells appear twice with different `who` slugs (`alvarez2022-1pts` and `alvarez2022-3pts`) | both files identical | Will dedup to single file in v1.1 |
|
| 267 |
+
| **Gene-space heterogeneity** — 18 distinct gene-space sizes across cohorts (Smart-seq2 ~54k vs 10x v3 ~36k vs reduced curation ~2-3k) | all multi-cohort concat operations | Use `ad.concat(..., join='outer', fill_value=0)`; foundation model fine-tune should project to common Ensembl space |
|
| 268 |
+
| **Some cohorts use HGNC symbol as var.index, others use Ensembl ID** | mixed across GEO vs CELLxGENE | Documented per-file in `uns.aivin_obs_field_notes`; v2 will unify to Ensembl ID |
|
| 269 |
+
|
| 270 |
+
---
|
| 271 |
+
|
| 272 |
+
## Contact
|
| 273 |
+
|
| 274 |
+
- 🤗 **HF discussions tab** on this repo (preferred for technical questions)
|
| 275 |
+
- 💬 **scverse Discourse**: https://discourse.scverse.org/ — `#show-and-tell` thread
|
| 276 |
+
- 📧 **z4fu@ucsd.edu** (project lead)
|
| 277 |
+
- 🐛 **Issues / PRs**: `github.com/AIVIN-UCSD/aivin`
|
| 278 |
+
|
| 279 |
+
---
|
| 280 |
+
|
| 281 |
+
*Last updated: 2026-05-25 · AIVIN v2.0 snapshot 2026-Q2 · 41 .h5ad (40 valid) · 1.17M cells + 35k spots · ~14 GB*
|
README.md.bak
ADDED
|
@@ -0,0 +1,281 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: cc-by-4.0
|
| 3 |
+
language:
|
| 4 |
+
- en
|
| 5 |
+
size_categories:
|
| 6 |
+
- 1M<n<10M
|
| 7 |
+
task_categories:
|
| 8 |
+
- feature-extraction
|
| 9 |
+
tags:
|
| 10 |
+
- single-cell
|
| 11 |
+
- scRNA-seq
|
| 12 |
+
- snRNA-seq
|
| 13 |
+
- spatial-transcriptomics
|
| 14 |
+
- visium
|
| 15 |
+
- cancer
|
| 16 |
+
- hepatocellular-carcinoma
|
| 17 |
+
- HCC
|
| 18 |
+
- iCCA
|
| 19 |
+
- liver
|
| 20 |
+
- atlas
|
| 21 |
+
- cellxgene
|
| 22 |
+
- anndata
|
| 23 |
+
pretty_name: AIVIN Liver References
|
| 24 |
+
---
|
| 25 |
+
|
| 26 |
+
# AIVIN · Liver References
|
| 27 |
+
|
| 28 |
+
|
| 29 |
+
|
| 30 |
+
Harmonized single-cell + single-nucleus + Visium spatial reference set for
|
| 31 |
+
**human liver tissue** (cancer + healthy + chronic-liver-disease), curated
|
| 32 |
+
under the AIVIN cross-cancer reference project at UCSD. All files are
|
| 33 |
+
[CELLxGENE schema 7.0.0](https://github.com/chanzuckerberg/single-cell-curation)
|
| 34 |
+
compatible and follow the unified AIVIN naming + provenance convention.
|
| 35 |
+
|
| 36 |
+
| | |
|
| 37 |
+
|---|---|
|
| 38 |
+
| **Files** | **41** `.h5ad` (40 valid + 1 known-broken — see Known Issues) |
|
| 39 |
+
| **Cells (sc + sn)** | **~1,177,000** |
|
| 40 |
+
| **Visium spots** | **~34,900** |
|
| 41 |
+
| **Distinct cohorts** | **18** (12 cancer + 5 healthy + 1 CLD) |
|
| 42 |
+
| **Source studies** | 15 GEO + CELLxGENE Census + 1 GSM-direct |
|
| 43 |
+
| **Disease dimensions** | HCC primary · HCC metastatic · HCC fetal · HCC anti-PD1 · HCC MASH · iCCA · cholangiocarcinoma · NASH · HCV · chronic liver disease · PSC · PBC · healthy |
|
| 44 |
+
| **Platforms** | 10x Chromium v2/v3 · Smart-seq2 · CEL-Seq2 · 10x Visium |
|
| 45 |
+
| **Total size** | ~14 GB |
|
| 46 |
+
| **Snapshot** | AIVIN 2026-Q2 v1.0 |
|
| 47 |
+
| **Zenodo DOI** | `[pending — Sat 5/30 snapshot]` |
|
| 48 |
+
| **HF DOI** | `[pending — mint after upload]` |
|
| 49 |
+
| **License** | CC-BY-4.0 (plus cite original cohort papers) |
|
| 50 |
+
|
| 51 |
+
---
|
| 52 |
+
|
| 53 |
+
## What's in this repo
|
| 54 |
+
|
| 55 |
+
Every `.h5ad` follows the AIVIN naming convention (see `NAMING_INDEX §八` in the AIVIN GitHub for the full grammar):
|
| 56 |
+
|
| 57 |
+
```
|
| 58 |
+
<modality>__<cohort-slug>__<who>__<NcxMg>__<accession>.h5ad
|
| 59 |
+
│ │ │ │ └── GEO GSE / GSM · GSA HRA · CELLxGENE UUID · Zenodo ID
|
| 60 |
+
│ │ │ └────────────── shape: cells × genes (or spots × genes for visium)
|
| 61 |
+
│ │ └────────────────────── first-author + year + n donors/samples
|
| 62 |
+
│ └─────────────────────────────────── biology tag (disease + sub-type)
|
| 63 |
+
└───────────────────────────────────────────── modality: sc = single-cell · sn = single-nucleus · visium = spatial
|
| 64 |
+
```
|
| 65 |
+
|
| 66 |
+
### Cohort manifest
|
| 67 |
+
|
| 68 |
+
#### Cancer cohorts (GEO source · 24 files · ~874k cells)
|
| 69 |
+
|
| 70 |
+
| Cohort slug | Source | Cells | Disease | Platform | Citation |
|
| 71 |
+
|---|---|---:|---|---|---|
|
| 72 |
+
| `hcc-cd45` | GSE235863 | 191,435 | HCC CD45+ enriched | 10x | Guo et al., 2025 |
|
| 73 |
+
| `hcc-fetal` | GSE156625 | 109,238 | HCC onco-fetal | 10x | Sharma et al., *Cell* 2020 |
|
| 74 |
+
| `hcc-cd8tcell` | GSE235863 | 95,408 | HCC CD8 T cells | 10x | Guo et al., 2025 |
|
| 75 |
+
| `hcc-tumor-normal` (Sharma) | GSE156625 | 73,589 | HCC tumor + adjacent | 10x | Sharma et al., *Cell* 2020 |
|
| 76 |
+
| `hcc-multisite` | GSE149614 | 71,915 | HCC primary + metastatic + PVTT + LN | 10x | Lu et al., *Nat Commun* 2022 |
|
| 77 |
+
| `hcc-iccA-cd45` | GSE140228-droplet | 66,187 | HCC + iCCA, CD45+ | 10x | Sharma et al., *Cell* 2020 |
|
| 78 |
+
| `hcc-iccA-treated` | GSE151530 | 56,721 | HCC + iCCA, post-treatment | 10x | Ma et al., *J Hepatol* 2021 |
|
| 79 |
+
| `hcc-trm` | GSE281110 | 41,848 | HCC tumor-associated TRM T | 10x | Park et al., 2025 |
|
| 80 |
+
| `hcc-tumor-normal-3pt` | GSE189175 | 39,995 | HCC tumor + normal | sn-10x | Alvarez et al., 2022 |
|
| 81 |
+
| `hcc-tumor-normal-1pt` | GSE189175 | 39,995 | (duplicate — see Known Issues) | sn-10x | Alvarez et al., 2022 |
|
| 82 |
+
| `hcc-mash-spectrum` | GSE282630 | 34,396 | HCC + MASH spectrum | 10x | Huang et al., 2025 |
|
| 83 |
+
| `hcc-cd45-ss2` | GSE140228-ss2 | 7,074 | HCC CD45+ Smart-seq2 | SS2 | Sharma et al., *Cell* 2020 |
|
| 84 |
+
| `hcc-iccA-mixed-set1` | GSE125449-set1 | 5,115 | HCC + iCCA, mixed | 10x | Ma et al., *Cancer Cell* 2019 |
|
| 85 |
+
| `hcc-tcell` | GSE98638 | 5,063 | HCC infiltrating T cells | SMART-seq2 | Zheng et al., *Cell* 2017 |
|
| 86 |
+
| `hcc-iccA-mixed-set2` | GSE125449-set2 | 4,831 | HCC + iCCA, mixed | 10x | Ma et al., *Cancer Cell* 2019 |
|
| 87 |
+
| `hcc-antiPD1` (R1) | GSE238264-HCC1R | 3,006 | HCC anti-PD1 responder | 10x | Liu et al., 2025 |
|
| 88 |
+
| `hcc-antiPD1` (R4) | GSE238264-HCC4R | 3,002 | HCC anti-PD1 responder | 10x | Liu et al., 2025 |
|
| 89 |
+
| `hcc-antiPD1` (R2) | GSE238264-HCC2R | 2,766 | HCC anti-PD1 responder | 10x | Liu et al., 2025 |
|
| 90 |
+
| `hcc-antiPD1` (NR6) | GSE238264-HCC6NR | 2,575 | HCC anti-PD1 non-responder | 10x | Liu et al., 2025 |
|
| 91 |
+
| `hcc-antiPD1` (NR7) | GSE238264-HCC7NR | 2,453 | HCC anti-PD1 non-responder | 10x | Liu et al., 2025 |
|
| 92 |
+
| `hcc-antiPD1` (R3) | GSE238264-HCC3R | 2,170 | HCC anti-PD1 responder | 10x | Liu et al., 2025 |
|
| 93 |
+
| `hcc-antiPD1` (NR5) | GSE238264-HCC5NR | 1,320 | HCC anti-PD1 non-responder | 10x | Liu et al., 2025 |
|
| 94 |
+
| `cld-lyec` | GSE129933 | 901 | Chronic liver disease lymphatic EC | SMART-seq2 | Tamburini et al., *Front Immunol* 2019 |
|
| 95 |
+
| `healthy-nat` | GSM4648565 | 13,083 | healthy liver | 10x | (Nat Commun 2020) |
|
| 96 |
+
|
| 97 |
+
#### Healthy + autoimmune baselines (CELLxGENE Census · 11 sc/sn files · ~303k cells)
|
| 98 |
+
|
| 99 |
+
| Cohort slug | Cells | Cell type / disease | Modality |
|
| 100 |
+
|---|---:|---|---|
|
| 101 |
+
| `psc-pbc-healthy` (sn) | 105,780 | PSC + PBC + healthy, all cells | sn |
|
| 102 |
+
| `psc-pbc-healthy` (sc) | 89,637 | PSC + PBC + healthy, all cells | sc |
|
| 103 |
+
| `healthy hepatocyte-v1` | 53,015 | hepatocytes | sc |
|
| 104 |
+
| `healthy lymphoid` | 16,665 | lymphoid lineage | sc |
|
| 105 |
+
| `healthy hepatocyte-v2` | 13,635 | hepatocytes (alt curation) | sc |
|
| 106 |
+
| `healthy macrophage` | 11,127 | macrophages | sc |
|
| 107 |
+
| `healthy endothelial` | 9,422 | endothelial cells | sc |
|
| 108 |
+
| `healthy stellate` | 1,417 | hepatic stellate cells | sc |
|
| 109 |
+
| `healthy b-cell` | 1,250 | B cells | sc |
|
| 110 |
+
| `healthy cholangiocyte` | 1,011 | cholangiocytes | sc |
|
| 111 |
+
|
| 112 |
+
#### Spatial transcriptomics (CELLxGENE Census · 6 Visium files · ~35k spots)
|
| 113 |
+
|
| 114 |
+
| Cohort slug | Spots | Tissue block | Disease |
|
| 115 |
+
|---|---:|---|---|
|
| 116 |
+
| `visium healthy-C73 / blockA1` | 4,992 | block A1 | healthy donor C73 |
|
| 117 |
+
| `visium healthy-C73 / blockC1` | 4,992 | block C1 | healthy donor C73 |
|
| 118 |
+
| `visium healthy-C73 / blockD1` | 4,992 | block D1 | healthy donor C73 |
|
| 119 |
+
| `visium psc-PSC011 / blockA1` | 4,992 | block A1 | PSC patient 011 |
|
| 120 |
+
| `visium psc-PSC011 / blockB1` | 4,992 | block B1 | PSC patient 011 |
|
| 121 |
+
| `visium psc-PSC011 / blockC1` | 4,992 | block C1 | PSC patient 011 |
|
| 122 |
+
| `visium psc-PSC011 / blockD1` | 4,992 | block D1 | PSC patient 011 |
|
| 123 |
+
|
| 124 |
+
---
|
| 125 |
+
|
| 126 |
+
## Schema
|
| 127 |
+
|
| 128 |
+
All `.h5ad` conform to **CELLxGENE schema 7.0.0** plus AIVIN extensions:
|
| 129 |
+
|
| 130 |
+
**`obs` (cells) — required columns**
|
| 131 |
+
|
| 132 |
+
- `cell_id` (index)
|
| 133 |
+
- `donor_id` (when known)
|
| 134 |
+
- `tissue_site` — unified vocab: `PT` (primary tumor) · `NTL` (normal liver) · `JTL` (juxta-tumor liver) · `MLN` (lymph node metastasis) · `PVTT` (portal vein tumor thrombus) · `PBMC` (peripheral blood) · `LIL` (liver intra-lesional)
|
| 135 |
+
- `disease` — values within the **Disease dimensions** list above
|
| 136 |
+
- `cell_type` (when annotated by original author)
|
| 137 |
+
- `assay` — platform / chemistry
|
| 138 |
+
|
| 139 |
+
**`var` (genes) — convention**
|
| 140 |
+
|
| 141 |
+
- Ensembl ID as `var.index` (when available, esp. CELLxGENE-sourced)
|
| 142 |
+
- Some GEO-sourced cohorts use HGNC `gene_symbol` as index + `entrez_id` column
|
| 143 |
+
- Heterogeneity across cohorts: 18 distinct gene-space sizes (2,384 – 58,100 genes) — see `aivin_obs_field_notes` per file for caveats; downstream concat use `ad.concat(..., join='outer')`
|
| 144 |
+
|
| 145 |
+
**`uns` (provenance, AIVIN-specific)**
|
| 146 |
+
|
| 147 |
+
- `citation` — full APA reference
|
| 148 |
+
- `doi` — primary paper DOI
|
| 149 |
+
- `source_accession` — GEO GSE / GSM / GSA HRA / CELLxGENE UUID / Zenodo ID
|
| 150 |
+
- `source_url`
|
| 151 |
+
- `aivin_ingest_date`
|
| 152 |
+
- `aivin_cohort_slug`
|
| 153 |
+
- `aivin_source_files` — original raw filename list
|
| 154 |
+
- `aivin_obs_field_notes` — any value-mapping done in ingest
|
| 155 |
+
|
| 156 |
+
---
|
| 157 |
+
|
| 158 |
+
## Usage
|
| 159 |
+
|
| 160 |
+
### Load one cohort (lazy / single file)
|
| 161 |
+
|
| 162 |
+
```python
|
| 163 |
+
from huggingface_hub import hf_hub_download
|
| 164 |
+
import anndata as ad
|
| 165 |
+
|
| 166 |
+
path = hf_hub_download(
|
| 167 |
+
repo_id='ucsd-aivin/liver-references',
|
| 168 |
+
filename='sc__hcc-multisite__lu2022-10pts__71915cx25712g__GSE149614.h5ad',
|
| 169 |
+
repo_type='dataset',
|
| 170 |
+
)
|
| 171 |
+
a = ad.read_h5ad(path)
|
| 172 |
+
print(a)
|
| 173 |
+
# Inspect AIVIN provenance
|
| 174 |
+
print(a.uns['citation'])
|
| 175 |
+
print(a.uns['aivin_obs_field_notes'])
|
| 176 |
+
```
|
| 177 |
+
|
| 178 |
+
### Load all cancer cohorts + concat (gene union)
|
| 179 |
+
|
| 180 |
+
```python
|
| 181 |
+
from huggingface_hub import snapshot_download
|
| 182 |
+
from pathlib import Path
|
| 183 |
+
import anndata as ad
|
| 184 |
+
|
| 185 |
+
local = snapshot_download(
|
| 186 |
+
repo_id='ucsd-aivin/liver-references',
|
| 187 |
+
repo_type='dataset',
|
| 188 |
+
allow_patterns='sc__hcc-*.h5ad', # cancer only
|
| 189 |
+
ignore_patterns='*macparland2019-0donors*', # skip known-broken file
|
| 190 |
+
)
|
| 191 |
+
adatas = {f.stem: ad.read_h5ad(f) for f in Path(local).glob('sc__hcc-*.h5ad')}
|
| 192 |
+
merged = ad.concat(adatas, axis=0, join='outer', label='cohort', fill_value=0)
|
| 193 |
+
print(merged)
|
| 194 |
+
# ~750k cells × union of genes across cohorts
|
| 195 |
+
```
|
| 196 |
+
|
| 197 |
+
### Pipe into scvi-tools (foundation model training)
|
| 198 |
+
|
| 199 |
+
```python
|
| 200 |
+
import scvi
|
| 201 |
+
scvi.model.SCVI.setup_anndata(merged, batch_key='cohort')
|
| 202 |
+
model = scvi.model.SCVI(merged, n_layers=2, n_latent=30)
|
| 203 |
+
model.train(accelerator='mps') # Apple Silicon MPS acceleration
|
| 204 |
+
```
|
| 205 |
+
|
| 206 |
+
---
|
| 207 |
+
|
| 208 |
+
## Citation
|
| 209 |
+
|
| 210 |
+
If you use this dataset in a publication, please cite:
|
| 211 |
+
|
| 212 |
+
1. **AIVIN as a collection** (this dataset card):
|
| 213 |
+
```bibtex
|
| 214 |
+
@dataset{aivin_liver_2026Q2,
|
| 215 |
+
author = {AIVIN Project, UCSD},
|
| 216 |
+
title = {{AIVIN Liver References (2026-Q2 v1.0)}},
|
| 217 |
+
year = {2026},
|
| 218 |
+
publisher = {Hugging Face},
|
| 219 |
+
doi = {[pending HF DOI mint]},
|
| 220 |
+
url = {https://huggingface.co/datasets/ucsd-aivin/liver-references}
|
| 221 |
+
}
|
| 222 |
+
```
|
| 223 |
+
|
| 224 |
+
2. **Each individual cohort** — see the `uns.citation` field of every `.h5ad`,
|
| 225 |
+
or the **Cohort manifest** table above. Particularly for landmark papers:
|
| 226 |
+
- Lu et al., *Nat Commun* 13:4594 (2022) — `doi:10.1038/s41467-022-32283-3`
|
| 227 |
+
- Sharma et al., *Cell* 183:377 (2020) — `doi:10.1016/j.cell.2020.08.040`
|
| 228 |
+
- Ma et al., *J Hepatol* 75:1418 (2021) — `doi:10.1016/j.jhep.2021.06.028`
|
| 229 |
+
- Ma et al., *Cancer Cell* 36:418 (2019) — `doi:10.1016/j.ccell.2019.08.007`
|
| 230 |
+
- Zheng et al., *Cell* 169:1342 (2017) — `doi:10.1016/j.cell.2017.05.035`
|
| 231 |
+
|
| 232 |
+
3. **(Optional) the Zenodo permanent snapshot** for byte-frozen reproducibility:
|
| 233 |
+
`doi: [pending Sat 5/30]`
|
| 234 |
+
|
| 235 |
+
---
|
| 236 |
+
|
| 237 |
+
## License
|
| 238 |
+
|
| 239 |
+
This collection is released under **CC-BY-4.0**. The license applies to AIVIN's
|
| 240 |
+
harmonization, schema mapping, and provenance metadata. **You must still cite
|
| 241 |
+
the original cohort papers** when using their data — see the per-cohort
|
| 242 |
+
manifest above. Cohorts derived from controlled-access sources (e.g., GSA-Human
|
| 243 |
+
HRA001748 Xue 2022) are NOT included in this public repo; see the cross-tissue
|
| 244 |
+
meta-repo for access pointers.
|
| 245 |
+
|
| 246 |
+
---
|
| 247 |
+
|
| 248 |
+
## Pipeline & reproducibility
|
| 249 |
+
|
| 250 |
+
- **Ingest scripts**: `github.com/ucsd-aivin/aivin/tree/main/scripts`
|
| 251 |
+
(per-cohort `<Cn>_<author><year>_ingest.py` + `W3_backlog_ingest.py` dispatcher)
|
| 252 |
+
- **Methods extracts**: per-paper structured methods at
|
| 253 |
+
`github.com/ucsd-aivin/aivin/tree/main/literature/A_cancer_TME/methods_extracts`
|
| 254 |
+
- **Structure report**: full per-file schema audit at
|
| 255 |
+
`github.com/ucsd-aivin/aivin/blob/main/database_unified/Liver_References/STRUCTURE_REPORT.md`
|
| 256 |
+
- **Backlog inventory**: candidates for v3 (3-month) expansion at
|
| 257 |
+
`github.com/ucsd-aivin/aivin/blob/main/database_unified/_staging/BACKLOG_INVENTORY.md`
|
| 258 |
+
|
| 259 |
+
---
|
| 260 |
+
|
| 261 |
+
## Known issues (v1.0)
|
| 262 |
+
|
| 263 |
+
| Issue | Affected file | Fix planned |
|
| 264 |
+
|---|---|---|
|
| 265 |
+
| **MacParland v1 ingest broken** (shape `0 × 3,958,008`) — the multi-plate CEL-Seq2 concat in `ingest_GSE124395()` produced a degenerate output | `sc__healthy-hlca__macparland2019-0donors__0cx3958008g__GSE124395.h5ad` | Will re-ingest in v1.1 with proper plate-level dedup; **filter out via `ignore_patterns='*macparland2019-0donors*'`** in `snapshot_download` |
|
| 266 |
+
| **GSE189175 Alvarez duplicate** — same 39,995 cells appear twice with different `who` slugs (`alvarez2022-1pts` and `alvarez2022-3pts`) | both files identical | Will dedup to single file in v1.1 |
|
| 267 |
+
| **Gene-space heterogeneity** — 18 distinct gene-space sizes across cohorts (Smart-seq2 ~54k vs 10x v3 ~36k vs reduced curation ~2-3k) | all multi-cohort concat operations | Use `ad.concat(..., join='outer', fill_value=0)`; foundation model fine-tune should project to common Ensembl space |
|
| 268 |
+
| **Some cohorts use HGNC symbol as var.index, others use Ensembl ID** | mixed across GEO vs CELLxGENE | Documented per-file in `uns.aivin_obs_field_notes`; v2 will unify to Ensembl ID |
|
| 269 |
+
|
| 270 |
+
---
|
| 271 |
+
|
| 272 |
+
## Contact
|
| 273 |
+
|
| 274 |
+
- 🤗 **HF discussions tab** on this repo (preferred for technical questions)
|
| 275 |
+
- 💬 **scverse Discourse**: https://discourse.scverse.org/ — `#show-and-tell` thread
|
| 276 |
+
- 📧 **z4fu@ucsd.edu** (project lead)
|
| 277 |
+
- 🐛 **Issues / PRs**: `github.com/ucsd-aivin/aivin`
|
| 278 |
+
|
| 279 |
+
---
|
| 280 |
+
|
| 281 |
+
*Last updated: 2026-05-25 · AIVIN v2.0 snapshot 2026-Q2 · 41 .h5ad (40 valid) · 1.17M cells + 35k spots · ~14 GB*
|
STRUCTURE_REPORT.md
ADDED
|
@@ -0,0 +1,245 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# database_unified/Liver_References · 数据结构与特征报告
|
| 2 |
+
|
| 3 |
+
**生成日期**：2026-05-17（v1 首次写入；与 Colorectal/Breast/Gastric 三份 STRUCTURE_REPORT 体例对齐）
|
| 4 |
+
**对象目录**：`database_unified/Liver_References/`（18 份 `.h5ad`，~6.96 GB）+ 计划加入 4 份 Wave 2 HCC scRNA dataset
|
| 5 |
+
**目的**：在做任何清洗 / 合并 / 派生视图之前，沉淀一份不削减、不假设用途的结构档案。本报告与 `Colorectal_References/STRUCTURE_REPORT.md` 共用同一份模板，便于跨脏器横向对照。
|
| 6 |
+
**附属机器可读资料**：
|
| 7 |
+
- `_provenance/uuid_to_new_name.csv` — 18 文件 UUID/GSM ↔ 新文件名双向映射
|
| 8 |
+
- `merged_views/MERGE_REPORT.md` — 子包内合并视图说明 + dashboard
|
| 9 |
+
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
## 一、来源与许可
|
| 13 |
+
|
| 14 |
+
当前 18 份 `.h5ad` 来自**两个独立来源**：
|
| 15 |
+
|
| 16 |
+
| 子包 | 文件数 | 来源 | DOI | License |
|
| 17 |
+
|---|---:|---|---|---|
|
| 18 |
+
| **A · 健康肝 per-cell-type 子集** | 8 | CZ CELLxGENE Discover（同一 collection） | `10.1016/j.jhep.2023.12.023` | CC-BY-4.0 |
|
| 19 |
+
| **B · 全细胞图谱（PSC + PBC + healthy）** | 2 | 同 A · CELLxGENE | 同 A | 同 A |
|
| 20 |
+
| **C · Visium 空间转录组** | 7 | 同 A · CELLxGENE | 同 A | 同 A |
|
| 21 |
+
| **D · GEO 增量 NAT raw counts** | 1 | GEO GSE153643（Regev lab, Broad Institute, 2021-05） | — | GEO 默认公开 |
|
| 22 |
+
|
| 23 |
+
身份信息汇总：
|
| 24 |
+
|
| 25 |
+
| 字段 | 值 |
|
| 26 |
+
|---|---|
|
| 27 |
+
| 数据来源（主） | CZ CELLxGENE Discover（A/B/C：1 个 collection · DOI `10.1016/j.jhep.2023.12.023`） |
|
| 28 |
+
| 数据来源（增量） | GEO GSE153643（D：Regev lab "Single cells from adipose and liver"） |
|
| 29 |
+
| 物种 | *Homo sapiens*（全部 18 份） |
|
| 30 |
+
| AnnData schema | A/B/C 全部 CZ CELLxGENE **7.0.0**；D 是 raw .h5ad（**几乎没有 obs 元数据**，see §gotcha） |
|
| 31 |
+
| 总细胞 / spot 数 | scRNA 195,417 + snRNA 105,780 + Visium spot 34,944 + GEO sc 13,083 = **349,224** |
|
| 32 |
+
| 独立供体数 | A/B/C：12 healthy + PSC011 + PBC003-PBC004 + healthy-C73 等 ~15 名；D：1 NAT donor · **跨子包无重叠** |
|
| 33 |
+
| 累计大小 | 约 **6.96 GB** |
|
| 34 |
+
| 唯一疾病维度 | **healthy / PSC / PBC / NAT**（**无 HCC / iCCA / 转移 / ICB**——这是 Wave 2 要补的差距） |
|
| 35 |
+
|
| 36 |
+
---
|
| 37 |
+
|
| 38 |
+
## 二、目录现状一览
|
| 39 |
+
|
| 40 |
+
物理布局：
|
| 41 |
+
|
| 42 |
+
```
|
| 43 |
+
database_unified/
|
| 44 |
+
└── Liver_References/ ← ~6.96 GB
|
| 45 |
+
├── sc__healthy__b-cell__1250cx32596g__<UUID>.h5ad ← 子包 A · 健康肝 per-cell-type × 8 文件
|
| 46 |
+
├── sc__healthy__cholangiocyte__1011cx32596g__<UUID>.h5ad
|
| 47 |
+
├── sc__healthy__endothelial__9422cx32596g__<UUID>.h5ad
|
| 48 |
+
├── sc__healthy__hepatocyte-v1__53015cx32596g__<UUID>.h5ad
|
| 49 |
+
├── sc__healthy__hepatocyte-v2__13635cx32596g__<UUID>.h5ad
|
| 50 |
+
├── sc__healthy__lymphoid__16665cx32596g__<UUID>.h5ad
|
| 51 |
+
├── sc__healthy__macrophage__11127cx32596g__<UUID>.h5ad
|
| 52 |
+
├── sc__healthy__stellate__1417cx32596g__<UUID>.h5ad
|
| 53 |
+
├── sc__psc-pbc-healthy__all-cells__89637cx32596g__<UUID>.h5ad ← 子包 B · 主图谱 × 2 文件
|
| 54 |
+
├── sn__psc-pbc-healthy__all-cells__105780cx32596g__<UUID>.h5ad
|
| 55 |
+
├── visium__healthy-C73__block{A1,C1,D1}__4992sx35477g__<UUID>.h5ad ← 子包 C · Visium × 7 文件
|
| 56 |
+
├── visium__psc-PSC011__block{A1,B1,C1,D1}__4992sx35477g__<UUID>.h5ad
|
| 57 |
+
├── sc__healthy-nat__liver__13083cx33694g__GSM4648565.h5ad ← 子包 D · GEO 增量 × 1 文件
|
| 58 |
+
├── AI_powered_CCRM_liver.pdf ← 配套文献（早期）
|
| 59 |
+
├── AI_powered_CCRM_liver__reading_notes.md ← 阅读笔记
|
| 60 |
+
├── merged_views/ ← 派生视图（合并子包）
|
| 61 |
+
│ ├── sc__healthy__merged-all-celltypes__107542cx32596g.h5ad
|
| 62 |
+
│ ├── visium__healthy-C73__merged-all-blocks__14976sx35477g.h5ad
|
| 63 |
+
│ ├── visium__psc-PSC011__merged-all-blocks__19968sx35477g.h5ad
|
| 64 |
+
│ ├── MERGE_REPORT.md
|
| 65 |
+
│ └── dashboard.html
|
| 66 |
+
└── _provenance/
|
| 67 |
+
└── uuid_to_new_name.csv ← 18 行双向映射
|
| 68 |
+
```
|
| 69 |
+
|
| 70 |
+
完整文件命名规则与对照表见 [`database_unified/NAMING_INDEX.md`](../NAMING_INDEX.md) §一-§七。
|
| 71 |
+
|
| 72 |
+
### 2.1 子包技术参数总览
|
| 73 |
+
|
| 74 |
+
| 子包 | Assay 平台 | 大小 | 细胞 / spot | 独立 donors | 基因数 | obsm 嵌入 | raw.X | 疾病维度 |
|
| 75 |
+
|---|---|---:|---:|---:|---:|---|:---:|---|
|
| 76 |
+
| **A · 健康肝 per-cell-type** | 10x 3'v3 主导（部分 5' tx） | ~2.5 GB | 107,542 | ~12 healthy | 32,596 | `X_umap`, `X_scvi`(some) | ✓ 8/8 | healthy only |
|
| 77 |
+
| **B · 全细胞主图谱** | 10x 3'v3 (sc) + 10x sn-3'v3 (sn) | ~3.0 GB | 195,417 | PSC011-PSC03 + PBC003-04 + healthy 对照 | 32,596 | `X_umap`, `X_scvi` | ✓ 2/2 | PSC + PBC + healthy |
|
| 78 |
+
| **C · Visium 空间** | Visium Spatial Gene Expression V1 | ~1.4 GB | 34,944 spots (7 × 4992) | healthy-C73 + psc-PSC011 各 1 donor | 35,477 | `spatial` only | ✗ 0/7 | healthy + PSC（�� HCC）|
|
| 79 |
+
| **D · GEO NAT raw** | 10x 3' v2 | 53 MB | 13,083 | 1 NAT donor (tumor-adjacent) | 33,694（**gene symbol 不是 Ensembl**） | ∅ | ✓ raw=True，**但矩阵转置**（X.attrs[shape] 是 [genes, cells]） | NAT（有 cancer field effect） |
|
| 80 |
+
| **合计 / 4 子包** | 4 种 assay | ~6.96 GB | **349,224** | ~15 donors | 多版本 | 异质 | 19/18（D 矩阵转置）| **4 种 disease level** |
|
| 81 |
+
|
| 82 |
+
**重要现状**：所有 18 文件**完全没有 HCC / iCCA / 肿瘤组织** — 当前 Liver_References 实际是"健康肝 + 慢性肝病（PSC/PBC）reference"，**HCC tumor 维度完全缺失**。这正是 Wave 2 的核心补丁。
|
| 83 |
+
|
| 84 |
+
---
|
| 85 |
+
|
| 86 |
+
## 三、每文件 schema 详细档案
|
| 87 |
+
|
| 88 |
+
由于本报告 v1 篇幅约束 + 信息已大量记录在 `NAMING_INDEX.md` 第二节改名对照表（18 行含中文 + UUID + cells × genes），本节给出**简化版**，不重复 NAMING_INDEX 已有的内容。
|
| 89 |
+
|
| 90 |
+
### 3.1 子包 A · 健康肝 per-cell-type 子集（8 份）
|
| 91 |
+
|
| 92 |
+
**Publication**：[`10.1016/j.jhep.2023.12.023`](https://doi.org/10.1016/j.jhep.2023.12.023)（Andrews et al.-style 健康肝多模态 atlas；CELLxGENE collection title 含 PSC + PBC + healthy 对照系列）
|
| 93 |
+
|
| 94 |
+
8 份 .h5ad 是健康肝的 **per-cell-type 切片**（b-cell / cholangiocyte / endothelial / hepatocyte-v1 / hepatocyte-v2 / lymphoid / macrophage / stellate），细胞数从 1,011（cholangiocyte）到 53,015（hepatocyte-v1）跨两个量级。所有 8 份共享：
|
| 95 |
+
|
| 96 |
+
- 32,596 基因（Ensembl IDs，schema-v7 var.feature_id）
|
| 97 |
+
- `X_umap` UMAP 坐标
|
| 98 |
+
- 部分含 `X_scvi` latent embedding（scvi-tools 整合时的产物）
|
| 99 |
+
- `raw.X = True`（保留 UMI count），下游可重做归一化
|
| 100 |
+
|
| 101 |
+
**Gotcha**：hepatocyte-v1（53k cells）vs hepatocyte-v2（13.6k cells）来自不同 sub-pipeline，**不是子集关系**——上游论文应该给了不同的处理粒度，本数据无法分辨"哪个更权威"。命名上 v1 / v2 中性，不抢决策（详见 NAMING_INDEX §1.2 第一条）。
|
| 102 |
+
|
| 103 |
+
### 3.2 子包 B · 全细胞主图谱（2 份）
|
| 104 |
+
|
| 105 |
+
- `sc__psc-pbc-healthy__all-cells__89637cx32596g__7d4d0da4-...` — scRNA-seq 全细胞，含 PSC + PBC + healthy 对照 89,637 cells
|
| 106 |
+
- `sn__psc-pbc-healthy__all-cells__105780cx32596g__4b5895d7-...` — snRNA-seq 全核，含同 cohort 105,780 nuclei
|
| 107 |
+
|
| 108 |
+
这两份是子包 A 的**反向视图**——A 是按 cell type 切片（每个文件单一 lineage），B 是按全细胞类型联合在一份 .h5ad 里（一个 master file 跨所有 lineage）。**两者细胞 ID 体系一致，可以做 sc/sn cross-modality 对照**。
|
| 109 |
+
|
| 110 |
+
### 3.3 子包 C · Visium 空间转录组（7 份）
|
| 111 |
+
|
| 112 |
+
7 份 Visium V1 切片来自两个 donor：
|
| 113 |
+
- **healthy-C73**（3 个 block：A1 / C1 / D1）——健康对照供体
|
| 114 |
+
- **psc-PSC011**（4 个 block：A1 / B1 / C1 / D1）——PSC 患者
|
| 115 |
+
|
| 116 |
+
每份 4,992 spots × 35,477 genes，`obsm['spatial']` 存切片坐标。**关键缺陷**：`raw.X = False`——下游做 spot deconvolution（cell2location / RCTD / Tangram）必须从 GEO / Space Ranger 原始 output 补 raw counts。
|
| 117 |
+
|
| 118 |
+
### 3.4 子包 D · GEO 增量 NAT（1 份）
|
| 119 |
+
|
| 120 |
+
`sc__healthy-nat__liver__13083cx33694g__GSM4648565.h5ad` — 来自 GSE153643（Regev lab，2021-05），是 HCC tumor-adjacent normal liver（NAT），不是真 healthy。**3 个 gotcha**（详见 NAMING_INDEX §7.3）：
|
| 121 |
+
|
| 122 |
+
1. **矩阵转置**：X.attrs[shape] = [33694, N]，obs/_index 装的是基因名，var/_index 装的是 barcode。读后必须 `adata = adata.T`
|
| 123 |
+
2. **obs / var 元信息几乎为零**：obs 只有 barcode，var 只有 gene symbol（不是 Ensembl），没有 cell_type / disease / donor_id / suspension_type / assay
|
| 124 |
+
3. **是 NAT 不是 healthy**：source 写明 "liver tissue tumor adjacent from liver resection"——有 cancer field effect 风险，**不能跟 healthy donor 数据混用**
|
| 125 |
+
|
| 126 |
+
---
|
| 127 |
+
|
| 128 |
+
## 四、Wave 2 计划补齐（HCC scRNA tumor 维度）
|
| 129 |
+
|
| 130 |
+
详见 [`_downloads/WAVE2_STATUS.md`](../_downloads/WAVE2_STATUS.md) 与 [`NAMING_INDEX.md`](../NAMING_INDEX.md) §八。当前进度：
|
| 131 |
+
|
| 132 |
+
| Wave 2 # | 数据 | accession | 状态（2026-05-17） | 预定文件名（落地后） |
|
| 133 |
+
|---|---|---|---|---|
|
| 134 |
+
| L1 | Lu 2022 HCC 4-site multisite | GSE149614 | raw staged 1.5 GB · ingest script ready · **需 user host 跑** | `sc__hcc-multisite__lu2022-10pts__71915cx25712g__GSE149614.h5ad` |
|
| 135 |
+
| L2 | Sharma 2020 HCC CD45+ Droplet | GSE140228-droplet | raw staged 308 MB（含 ss2 分支）· ingest 未启动 | `sc__hcc-cd45__sharma2020-droplet-16pts__NcxMg__GSE140228-droplet.h5ad` |
|
| 136 |
+
| L2b | Sharma 2020 同 series Smart-seq2 | GSE140228-ss2 | 同 L2 staged 53 MB csv | `sc__hcc-cd45__sharma2020-ss2-16pts__NcxMg__GSE140228-ss2.h5ad` |
|
| 137 |
+
| L8 | Ma 2021 HCC+iCCA treatment | GSE151530 | raw staged 292 MB · ingest 未启动 | `sc__hcc-iccA-treated__ma2021-37pts__NcxMg__GSE151530.h5ad` |
|
| 138 |
+
| L7 | Xue 2022 HCC TIME 124pts | HRA001748 | ⛔ DAC apply pending | `sc__hcc-time__xue2022-124pts__~1Mcx~20kg__HRA001748.h5ad` |
|
| 139 |
+
|
| 140 |
+
Wave 2 全落地后 Liver_References 将从**当前 18 个健康/PSC/PBC 文件**扩展为 **23 个文件**（增 4-5 �� HCC scRNA），覆盖 healthy / PSC / PBC / NAT / HCC multisite / HCC immune CD45+ / HCC+iCCA treatment / HCC TIME 共 8 个疾病维度。
|
| 141 |
+
|
| 142 |
+
---
|
| 143 |
+
|
| 144 |
+
## 五、命名规范
|
| 145 |
+
|
| 146 |
+
完整命名规则、改名对照表、GEO/GSA 增量约定见：[`database_unified/NAMING_INDEX.md`](../NAMING_INDEX.md)。
|
| 147 |
+
|
| 148 |
+
简要：`<modality>__<cohort>__<content>__<n>cx<m>g__<id>.h5ad`
|
| 149 |
+
- modality: `sc` / `sn` / `visium`
|
| 150 |
+
- cohort 词汇库（详见 NAMING_INDEX §1.1 + §8.1）：包含 `healthy` / `psc-pbc-healthy` / `healthy-C73` / `psc-PSC011` / `healthy-nat` / `hcc-multisite` / `hcc-cd45` / `hcc-iccA` / `hcc-treated` / `hcc-time` 等
|
| 151 |
+
- id: CELLxGENE UUID（36 字符） OR GEO GSM/GSE OR GSA HRA accession
|
| 152 |
+
|
| 153 |
+
---
|
| 154 |
+
|
| 155 |
+
## 六、Gotchas（下游使用前必读）
|
| 156 |
+
|
| 157 |
+
1. **GSE153643 sample 矩阵转置**（详见 §3.4）——D 子包 1 份 .h5ad 读后必须 `adata.T`
|
| 158 |
+
2. **Visium raw.X 缺失**——C 子包 7 份做 deconvolution 必须从 GEO 补 raw counts
|
| 159 |
+
3. **Adipose 已拆出独立目录**——GSE153643 同 series 的 `GSM4648564_adipose` 在 `database_unified/Adipose_References/`，不在本目录（避免跨组织污染）
|
| 160 |
+
4. **iCCA 驼峰特例**——cohort 槽内 `iccA` 是 §NAMING_INDEX 8.4 的精确例外，其他字段仍全小写
|
| 161 |
+
5. **当前无 HCC tumor 数据**——任何"HCC 相关 reference"使用前，先检查是否需等 Wave 2 L1 / L2 / L8 落地
|
| 162 |
+
6. **C73 是 healthy donor，PSC011 是 PSC 患者**——Visium 文件名内编码了 donor 身份，分析时不要混用
|
| 163 |
+
|
| 164 |
+
---
|
| 165 |
+
|
| 166 |
+
## 七、横向对照
|
| 167 |
+
|
| 168 |
+
| 脏器 References | 文件数 | 总细胞/spot | 子包数 | 疾病覆盖 | HCC tumor? |
|
| 169 |
+
|---|---:|---:|---:|---|:---:|
|
| 170 |
+
| **Liver**（本目录） | 18（+ 4-5 Wave 2 计划） | 349,224 | 4（+ Wave 2 计划） | healthy / PSC / PBC / NAT | ❌（Wave 2 后 ✅）|
|
| 171 |
+
| **Colorectal** | 34 | 4,082,955 | 4 | healthy / IBD / CRC / CRC-Met | n/a (CRC tumor ✅) |
|
| 172 |
+
| **Breast** | 38 | (TBD) | 4+ | TNBC / MBC / Pan-cancer TME | (Breast tumor ✅) |
|
| 173 |
+
| **Gastric** | 12 | (TBD) | 3 | healthy / GC / iCCA-mixed | (Gastric tumor ✅) |
|
| 174 |
+
|
| 175 |
+
完整跨脏器横向对照见各自 STRUCTURE_REPORT，跨脏器 DOWNLOAD 优先级见 [`_provenance/db_search_reports/db_search__{liver,breast,crc,gastric}.md`](../_provenance/db_search_reports/) 4 份子报告。
|
| 176 |
+
|
| 177 |
+
---
|
| 178 |
+
|
| 179 |
+
*v1 生成于 2026-05-17. 触发：Wave 2 启动 + 补齐 STRUCTURE_REPORT gap（之前 4 个 References 中仅 Liver 缺）.*
|
| 180 |
+
*下次更新触发：L1 ingest 落地（host run） + L2/L8 ingest 落地（沙盒可跑）.*
|
| 181 |
+
|
| 182 |
+
---
|
| 183 |
+
|
| 184 |
+
## 九、Wave 1+2 ingest 实际完成（2026-05-17 同会话）
|
| 185 |
+
|
| 186 |
+
本会话沙盒里实际完成了 5 个 dataset ingest（C1 落到 Colorectal_References，其余 4 个落到本目录）。Liver_References 从 18 → 23 文件，新增 5 个 HCC/iCCA tumor reference——**这是 AIVIN Liver 第一次有真正的肿瘤数据**。
|
| 187 |
+
|
| 188 |
+
### 9.1 新增 5 个文件（**3 个新子包 E / F / G**）
|
| 189 |
+
|
| 190 |
+
| 子包 | 文件 | 大小 | shape | cells × genes | donors | cohort 字段值 |
|
| 191 |
+
|---|---|---:|---|---:|---:|---|
|
| 192 |
+
| **E · Ma 2019 HCC/iCCA mixed Cancer Cell** | `sc__hcc-iccA-mixed__ma2019-set1-12pts__5115cx20124g__GSE125449-set1.h5ad` | 22 MB | 5,115 × 20,124 | 5,115 cells | 12 patients (P01-P13 missing P08) | `hcc-iccA-mixed` |
|
| 193 |
+
| **E · 同子包** | `sc__hcc-iccA-mixed__ma2019-set2-7pts__4831cx19572g__GSE125449-set2.h5ad` | 16 MB | 4,831 × 19,572 | 4,831 cells | 7 patients | `hcc-iccA-mixed` |
|
| 194 |
+
| **F · Ma 2021 HCC/iCCA treatment J Hepatol** | `sc__hcc-iccA-treated__ma2021-46pts__56721cx18667g__GSE151530.h5ad` | 230 MB | 56,721 × 18,667 | 56,721 cells | 46 sample_ids（37 paper patients，含 16 paired pre/post） | `hcc-iccA-treated` |
|
| 195 |
+
| **G · Sharma 2020 HCC CD45+ Cell** | `sc__hcc-iccA-cd45__sharma2020-droplet-5pts__66187cx54574g__GSE140228-droplet.h5ad` | 192 MB | 66,187 × 54,574 | 66,187 cells | 5 donors | `hcc-iccA-cd45`（**实测 5 donors HCC + CC mixed**，db_search 写的 "16 patients" 不准确 — 16 是含 ss2 sub-cohort 加 fetal 的总数）|
|
| 196 |
+
| **G · 同子包 ss2 分支** | `sc__hcc-cd45-ss2__sharma2020-ss2-6pts__7074cx54574g__GSE140228-ss2.h5ad` | 90 MB | 7,074 × 54,574 | 7,074 cells | 6 donors（与 droplet 部分重叠：D20171109） | `hcc-cd45-ss2` |
|
| 197 |
+
|
| 198 |
+
### 9.2 子包对照（**v1 vs v2**）
|
| 199 |
+
|
| 200 |
+
| 子包 | 文件数 v1 → v2 | cells v1 → v2 | 新增内容 |
|
| 201 |
+
|---|:---:|:---:|---|
|
| 202 |
+
| A · 健康肝 per-cell-type | 8 → 8 | 107,542 → 107,542 | 无变化 |
|
| 203 |
+
| B · 全细胞主图谱 | 2 → 2 | 195,417 → 195,417 | 无变化 |
|
| 204 |
+
| C · Visium 空间 | 7 → 7 | 34,944 → 34,944 | 无变化 |
|
| 205 |
+
| D · GEO NAT raw | 1 → 1 | 13,083 → 13,083 | 无变化 |
|
| 206 |
+
| **E · Ma 2019 HCC/iCCA mixed** | 0 → **2** | 0 → **9,946** | 新增（HCC + iCCA 共 cohort）|
|
| 207 |
+
| **F · Ma 2021 HCC/iCCA treatment** | 0 → **1** | 0 → **56,721** | 新增（含治疗前后 paired sub-cohort）|
|
| 208 |
+
| **G · Sharma 2020 HCC CD45+** | 0 → **2** | 0 → **73,261** | 新增（droplet + ss2 双平台，CD45+ 免疫富集）|
|
| 209 |
+
| **合计** | 18 → **23** | 349,224 → **489,152** | +5 文件 +139,928 cells |
|
| 210 |
+
|
| 211 |
+
### 9.3 疾病维���扩展
|
| 212 |
+
|
| 213 |
+
| v1 (2026-05-15) | v2 (2026-05-17) |
|
| 214 |
+
|---|---|
|
| 215 |
+
| healthy, PSC, PBC, NAT | healthy, PSC, PBC, NAT, **HCC, iCCA, HCC-CC mixed, HCC pre/post treatment** |
|
| 216 |
+
|
| 217 |
+
**最关键转变**：Liver_References 终于有了真正的肿瘤参考数据。下游做 cancer vs normal differential、TME 解析、tumor heterogeneity 都可以基于本目录内 self-contained 完成。
|
| 218 |
+
|
| 219 |
+
### 9.4 命名规范使用的新 cohort 词
|
| 220 |
+
|
| 221 |
+
本会话引入了 4 个新 cohort 词（已写入 §八 8.1 cohort 词汇库）：
|
| 222 |
+
- `hcc-iccA-mixed` — HCC + iCCA 同 cohort 无治疗信息（Ma 2019）
|
| 223 |
+
- `hcc-iccA-treated` — HCC + iCCA + 治疗前后维度（Ma 2021）
|
| 224 |
+
- `hcc-iccA-cd45` — HCC + iCCA + CD45+ 富集（Sharma droplet）
|
| 225 |
+
- `hcc-cd45-ss2` — HCC + CD45+ 富集 + Smart-seq2 平台（Sharma ss2）
|
| 226 |
+
|
| 227 |
+
iCCA 驼峰特例（§八 8.4 gotcha #3）在 3 个新 cohort 词里被使用，与 §一 全小写原则的精确例外是一致的。
|
| 228 |
+
|
| 229 |
+
### 9.5 待 PDF / DAC 后的 v2 enrichment
|
| 230 |
+
|
| 231 |
+
| 文件 | 缺什么 v1 字段 | 哪里来 |
|
| 232 |
+
|---|---|---|
|
| 233 |
+
| L3 Set1/Set2 | `cancer_type` (HCC vs iCCA per patient_code LCP##) | Ma 2019 *Cancer Cell* paper Supplementary Table 1 |
|
| 234 |
+
| L8 Ma 2021 | `cancer_type` (HCC vs iCCA), `timepoint` (pre/post), `treatment_type`, `treatment_response` | Ma 2021 *J Hepatol* paper Supplementary Table 1 |
|
| 235 |
+
| L2 / L2b | （fetal liver sub-cohort 在哪里？）| Sharma 2020 *Cell* paper 全文 + 可能 sample 名字解码 |
|
| 236 |
+
|
| 237 |
+
详见 `_downloads/WAVE2_STATUS.md` §1.2 用户下一步操作清单 + `literature/A_cancer_TME/methods_extracts/{12,13}_*.md` 的 v2 enrichment 字段清单。
|
| 238 |
+
|
| 239 |
+
### 9.6 还没 ingest 的：L1 Lu 2022
|
| 240 |
+
|
| 241 |
+
L1 Lu 2022 HCC multisite (71,915 cells × 25,712 genes, 4 sites NTL/PT/PVTT/MLN) **沙盒 ingest 失败**（dense 71,915-col matrix pandas chunked read 超 45s/chunk × 32 chunks），已写 host-runnable script `scripts/L1_lu2022_ingest.py`，user 在 Mac 跑 3-8 分钟即可落地。
|
| 242 |
+
|
| 243 |
+
---
|
| 244 |
+
|
| 245 |
+
*v2 生成于 2026-05-17 · 沙盒同会话完成 5/6 P0 dataset ingest（4 个落 Liver + 1 个落 Colorectal）· Liver 子包数 4 → 7 · cells 349k → 489k · 疾病维度 4 类 → 8 类。L1 host script + L7 DAC pending。*
|
_audit_report.md
ADDED
|
@@ -0,0 +1,55 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# AIVIN h5ad Audit Report
|
| 2 |
+
|
| 3 |
+
**Generated**: 2026-05-25T23:53:29.045338+00:00
|
| 4 |
+
|
| 5 |
+
**Files audited**: 40
|
| 6 |
+
|
| 7 |
+
**Pass**: 22 / 40
|
| 8 |
+
|
| 9 |
+
**View-only**: 3 / 40
|
| 10 |
+
|
| 11 |
+
|
| 12 |
+
## Per-file summary
|
| 13 |
+
|
| 14 |
+
| File | Pass | Shape | X state | uns OK | view-only | Notes |
|
| 15 |
+
|---|---|---|---|---|---|---|
|
| 16 |
+
| `sc__cld-lyec__tamburini2019-4pts__901cx11302g__GSE129933.h5ad` | ✓ | 901×11302 | raw_counts | 0/9 | | no patient/donor obs column found (Patient/patient_id/donor/subject); uns missing required |
|
| 17 |
+
| `sc__hcc-cd45-ss2__sharma2020-ss2-6pts__7074cx54574g__GSE140228-ss2.h5ad` | ✓ | 7074×54574 | unknown | 0/9 | 👁 | no patient/donor obs column found (Patient/patient_id/donor/subject); X state unknown — re |
|
| 18 |
+
| `sc__hcc-cd45__guo2025-9pts__191435cx28671g__GSE235863-9pt.h5ad` | ✓ | 191435×28671 | raw_counts | 12/9 | | |
|
| 19 |
+
| `sc__hcc-cd8tcell__guo2025-5pts__95408cx26256g__GSE235863-5pt.h5ad` | ✓ | 95408×26256 | raw_counts | 12/9 | | |
|
| 20 |
+
| `sc__hcc-fetal__sharma2020-62pts__109238cx2384g__GSE156625-HCCF.h5ad` | ✓ | 109238×2384 | scaled_HVG | 13/9 | 👁 | |
|
| 21 |
+
| `sc__hcc-iccA-cd45__sharma2020-droplet-5pts__66187cx54574g__GSE140228-droplet.h5ad` | ✓ | 66187×54574 | raw_counts | 0/9 | | no patient/donor obs column found (Patient/patient_id/donor/subject); uns missing required |
|
| 22 |
+
| `sc__hcc-iccA-mixed__ma2019-set1-12pts__5115cx20124g__GSE125449-set1.h5ad` | ✓ | 5115×20124 | raw_counts | 0/9 | | uns missing required keys: ['cohort_tag', 'ingest_date', 'ingested_by', 'modality', 'sourc |
|
| 23 |
+
| `sc__hcc-iccA-mixed__ma2019-set2-7pts__4831cx19572g__GSE125449-set2.h5ad` | ✓ | 4831×19572 | raw_counts | 0/9 | | uns missing required keys: ['cohort_tag', 'ingest_date', 'ingested_by', 'modality', 'sourc |
|
| 24 |
+
| `sc__hcc-iccA-treated__ma2021-46pts__56721cx18667g__GSE151530.h5ad` | ✓ | 56721×18667 | raw_counts | 0/9 | | no patient/donor obs column found (Patient/patient_id/donor/subject); uns missing required |
|
| 25 |
+
| `sc__hcc-mash-spectrum__huang2025-4pts__34396cx32285g__GSE282630.h5ad` | ✓ | 34396×32285 | raw_counts | 10/9 | | no patient/donor obs column found (Patient/patient_id/donor/subject); uns missing required |
|
| 26 |
+
| `sc__hcc-multisite__lu2022-10pts__71915cx25712g__GSE149614.h5ad` | ✓ | 71915×25712 | raw_counts | 0/9 | | uns missing required keys: ['cohort_tag', 'ingest_date', 'ingested_by', 'modality', 'sourc |
|
| 27 |
+
| `sc__hcc-trm__park2025-7pts__41848cx36601g__GSE281110.h5ad` | ✓ | 41848×36601 | raw_counts | 10/9 | | patient count mismatch: filename=7, obs.patient=6; uns missing required keys: ['x_state'] |
|
| 28 |
+
| `sc__hcc-tumor-normal__sharma2020-58pts__73589cx2608g__GSE156625-HCC.h5ad` | ✓ | 73589×2608 | scaled_HVG | 13/9 | 👁 | |
|
| 29 |
+
| `sc__healthy-nat__liver__13083cx33694g__GSM4648565.h5ad` | ✗ | ?×? | ? | 0/9 | | filename does not match AIVIN convention: sc__healthy-nat__liver__13083cx33694g__GSM464856 |
|
| 30 |
+
| `sc__healthy__b-cell__1250cx32596g__6ade4ff5-368c-4276-b051-818dc954da6d.h5ad` | ✗ | ?×? | ? | 0/9 | | filename does not match AIVIN convention: sc__healthy__b-cell__1250cx32596g__6ade4ff5-368c |
|
| 31 |
+
| `sc__healthy__cholangiocyte__1011cx32596g__601ef580-74ce-4a87-96e8-8e22bf4ed9fa.h5ad` | ✗ | ?×? | ? | 0/9 | | filename does not match AIVIN convention: sc__healthy__cholangiocyte__1011cx32596g__601ef5 |
|
| 32 |
+
| `sc__healthy__endothelial__9422cx32596g__a95e1659-9a48-4c55-8062-621ee4df9160.h5ad` | ✗ | ?×? | ? | 0/9 | | filename does not match AIVIN convention: sc__healthy__endothelial__9422cx32596g__a95e1659 |
|
| 33 |
+
| `sc__healthy__hepatocyte-v1__53015cx32596g__63137dcf-6236-464d-9018-e58c9323f59c.h5ad` | ✗ | ?×? | ? | 0/9 | | filename does not match AIVIN convention: sc__healthy__hepatocyte-v1__53015cx32596g__63137 |
|
| 34 |
+
| `sc__healthy__hepatocyte-v2__13635cx32596g__6b241dde-25c0-4edc-9021-03603ec3a524.h5ad` | ✗ | ?×? | ? | 0/9 | | filename does not match AIVIN convention: sc__healthy__hepatocyte-v2__13635cx32596g__6b241 |
|
| 35 |
+
| `sc__healthy__lymphoid__16665cx32596g__9bc7506b-7e73-4cea-bbb3-3603a016fbca.h5ad` | ✗ | ?×? | ? | 0/9 | | filename does not match AIVIN convention: sc__healthy__lymphoid__16665cx32596g__9bc7506b-7 |
|
| 36 |
+
| `sc__healthy__macrophage__11127cx32596g__2cfec927-9163-4684-ae04-c15175a6d781.h5ad` | ✗ | ?×? | ? | 0/9 | | filename does not match AIVIN convention: sc__healthy__macrophage__11127cx32596g__2cfec927 |
|
| 37 |
+
| `sc__healthy__stellate__1417cx32596g__08a9f031-1e76-405a-8050-0635743ce187.h5ad` | ✗ | ?×? | ? | 0/9 | | filename does not match AIVIN convention: sc__healthy__stellate__1417cx32596g__08a9f031-1e |
|
| 38 |
+
| `sc__psc-pbc-healthy__all-cells__89637cx32596g__7d4d0da4-655e-438a-a2ec-b4371e2b80fc.h5ad` | ✗ | ?×? | ? | 0/9 | | filename does not match AIVIN convention: sc__psc-pbc-healthy__all-cells__89637cx32596g__7 |
|
| 39 |
+
| `sn__hcc-tumor-normal__alvarez2022-1pts__39995cx58100g__GSE189175.h5ad` | ✓ | 39995×58100 | raw_counts | 10/9 | | var_names not unique: 1748 duplicates; patient count mismatch: filename=1, obs.Patient=3; |
|
| 40 |
+
| `sn__hcc-tumor-normal__alvarez2022-3pts__39995cx58100g__GSE189175.h5ad` | ✓ | 39995×58100 | raw_counts | 13/9 | | |
|
| 41 |
+
| `sn__psc-pbc-healthy__all-cells__105780cx32596g__4b5895d7-6d92-471a-b13a-5c59a000ddc4.h5ad` | ✗ | ?×? | ? | 0/9 | | filename does not match AIVIN convention: sn__psc-pbc-healthy__all-cells__105780cx32596g__ |
|
| 42 |
+
| `visium__hcc-antiPD1__liu2025-1pts__1320cx36601g__GSE238264-HCC5NR.h5ad` | ✓ | 1320×36601 | raw_counts | 10/9 | | no patient/donor obs column found (Patient/patient_id/donor/subject); uns missing required |
|
| 43 |
+
| `visium__hcc-antiPD1__liu2025-1pts__2170cx36601g__GSE238264-HCC3R.h5ad` | ✓ | 2170×36601 | raw_counts | 10/9 | | no patient/donor obs column found (Patient/patient_id/donor/subject); uns missing required |
|
| 44 |
+
| `visium__hcc-antiPD1__liu2025-1pts__2453cx36601g__GSE238264-HCC7NR.h5ad` | ✓ | 2453×36601 | raw_counts | 10/9 | | no patient/donor obs column found (Patient/patient_id/donor/subject); uns missing required |
|
| 45 |
+
| `visium__hcc-antiPD1__liu2025-1pts__2575cx36601g__GSE238264-HCC6NR.h5ad` | ✓ | 2575×36601 | raw_counts | 10/9 | | no patient/donor obs column found (Patient/patient_id/donor/subject); uns missing required |
|
| 46 |
+
| `visium__hcc-antiPD1__liu2025-1pts__2766cx36601g__GSE238264-HCC2R.h5ad` | ✓ | 2766×36601 | raw_counts | 10/9 | | no patient/donor obs column found (Patient/patient_id/donor/subject); uns missing required |
|
| 47 |
+
| `visium__hcc-antiPD1__liu2025-1pts__3002cx36601g__GSE238264-HCC4R.h5ad` | ✓ | 3002×36601 | raw_counts | 10/9 | | no patient/donor obs column found (Patient/patient_id/donor/subject); uns missing required |
|
| 48 |
+
| `visium__hcc-antiPD1__liu2025-1pts__3006cx36601g__GSE238264-HCC1R.h5ad` | ✓ | 3006×36601 | raw_counts | 10/9 | | no patient/donor obs column found (Patient/patient_id/donor/subject); uns missing required |
|
| 49 |
+
| `visium__healthy-C73__blockA1__4992sx35477g__9c731db0-9f44-4f8c-8139-9b9af2bcc782.h5ad` | ✗ | ?×? | ? | 0/9 | | filename does not match AIVIN convention: visium__healthy-C73__blockA1__4992sx35477g__9c73 |
|
| 50 |
+
| `visium__healthy-C73__blockC1__4992sx35477g__b2287ef1-eac3-49cc-93de-65df74e26a61.h5ad` | ✗ | ?×? | ? | 0/9 | | filename does not match AIVIN convention: visium__healthy-C73__blockC1__4992sx35477g__b228 |
|
| 51 |
+
| `visium__healthy-C73__blockD1__4992sx35477g__44453de7-1d66-4bd1-a83d-0b73a1690d57.h5ad` | ✗ | ?×? | ? | 0/9 | | filename does not match AIVIN convention: visium__healthy-C73__blockD1__4992sx35477g__4445 |
|
| 52 |
+
| `visium__psc-PSC011__blockA1__4992sx35477g__7bc39166-b664-4c92-922d-f9a8047768d2.h5ad` | ✗ | ?×? | ? | 0/9 | | filename does not match AIVIN convention: visium__psc-PSC011__blockA1__4992sx35477g__7bc39 |
|
| 53 |
+
| `visium__psc-PSC011__blockB1__4992sx35477g__6da751d4-63af-43b8-96db-395ca73dfb5f.h5ad` | ✗ | ?×? | ? | 0/9 | | filename does not match AIVIN convention: visium__psc-PSC011__blockB1__4992sx35477g__6da75 |
|
| 54 |
+
| `visium__psc-PSC011__blockC1__4992sx35477g__0cc59004-2b35-4767-8278-83e097ef32d1.h5ad` | ✗ | ?×? | ? | 0/9 | | filename does not match AIVIN convention: visium__psc-PSC011__blockC1__4992sx35477g__0cc59 |
|
| 55 |
+
| `visium__psc-PSC011__blockD1__4992sx35477g__721bfc1c-f77f-4c5a-afc6-04e3e3c675d3.h5ad` | ✗ | ?×? | ? | 0/9 | | filename does not match AIVIN convention: visium__psc-PSC011__blockD1__4992sx35477g__721bf |
|
_provenance/uuid_to_new_name.csv
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
old_uuid_filename,new_filename,size_mb,uuid_full,uuid8,cellxgene_dataset_url
|
| 2 |
+
08a9f031-1e76-405a-8050-0635743ce187.h5ad,sc__healthy__stellate__1417cx32596g__08a9f031-1e76-405a-8050-0635743ce187.h5ad,14.7,08a9f031-1e76-405a-8050-0635743ce187,08a9f031,https://datasets.cellxgene.cziscience.com/08a9f031-1e76-405a-8050-0635743ce187.h5ad
|
| 3 |
+
0cc59004-2b35-4767-8278-83e097ef32d1.h5ad,visium__psc-PSC011__blockC1__4992sx35477g__0cc59004-2b35-4767-8278-83e097ef32d1.h5ad,53.3,0cc59004-2b35-4767-8278-83e097ef32d1,0cc59004,https://datasets.cellxgene.cziscience.com/0cc59004-2b35-4767-8278-83e097ef32d1.h5ad
|
| 4 |
+
2cfec927-9163-4684-ae04-c15175a6d781.h5ad,sc__healthy__macrophage__11127cx32596g__2cfec927-9163-4684-ae04-c15175a6d781.h5ad,93.7,2cfec927-9163-4684-ae04-c15175a6d781,2cfec927,https://datasets.cellxgene.cziscience.com/2cfec927-9163-4684-ae04-c15175a6d781.h5ad
|
| 5 |
+
44453de7-1d66-4bd1-a83d-0b73a1690d57.h5ad,visium__healthy-C73__blockD1__4992sx35477g__44453de7-1d66-4bd1-a83d-0b73a1690d57.h5ad,1096.9,44453de7-1d66-4bd1-a83d-0b73a1690d57,44453de7,https://datasets.cellxgene.cziscience.com/44453de7-1d66-4bd1-a83d-0b73a1690d57.h5ad
|
| 6 |
+
4b5895d7-6d92-471a-b13a-5c59a000ddc4.h5ad,sn__psc-pbc-healthy__all-cells__105780cx32596g__4b5895d7-6d92-471a-b13a-5c59a000ddc4.h5ad,1543.8,4b5895d7-6d92-471a-b13a-5c59a000ddc4,4b5895d7,https://datasets.cellxgene.cziscience.com/4b5895d7-6d92-471a-b13a-5c59a000ddc4.h5ad
|
| 7 |
+
601ef580-74ce-4a87-96e8-8e22bf4ed9fa.h5ad,sc__healthy__cholangiocyte__1011cx32596g__601ef580-74ce-4a87-96e8-8e22bf4ed9fa.h5ad,16.7,601ef580-74ce-4a87-96e8-8e22bf4ed9fa,601ef580,https://datasets.cellxgene.cziscience.com/601ef580-74ce-4a87-96e8-8e22bf4ed9fa.h5ad
|
| 8 |
+
63137dcf-6236-464d-9018-e58c9323f59c.h5ad,sc__healthy__hepatocyte-v1__53015cx32596g__63137dcf-6236-464d-9018-e58c9323f59c.h5ad,439.8,63137dcf-6236-464d-9018-e58c9323f59c,63137dcf,https://datasets.cellxgene.cziscience.com/63137dcf-6236-464d-9018-e58c9323f59c.h5ad
|
| 9 |
+
6ade4ff5-368c-4276-b051-818dc954da6d.h5ad,sc__healthy__b-cell__1250cx32596g__6ade4ff5-368c-4276-b051-818dc954da6d.h5ad,18.1,6ade4ff5-368c-4276-b051-818dc954da6d,6ade4ff5,https://datasets.cellxgene.cziscience.com/6ade4ff5-368c-4276-b051-818dc954da6d.h5ad
|
| 10 |
+
6b241dde-25c0-4edc-9021-03603ec3a524.h5ad,sc__healthy__hepatocyte-v2__13635cx32596g__6b241dde-25c0-4edc-9021-03603ec3a524.h5ad,85.1,6b241dde-25c0-4edc-9021-03603ec3a524,6b241dde,https://datasets.cellxgene.cziscience.com/6b241dde-25c0-4edc-9021-03603ec3a524.h5ad
|
| 11 |
+
6da751d4-63af-43b8-96db-395ca73dfb5f.h5ad,visium__psc-PSC011__blockB1__4992sx35477g__6da751d4-63af-43b8-96db-395ca73dfb5f.h5ad,52.3,6da751d4-63af-43b8-96db-395ca73dfb5f,6da751d4,https://datasets.cellxgene.cziscience.com/6da751d4-63af-43b8-96db-395ca73dfb5f.h5ad
|
| 12 |
+
721bfc1c-f77f-4c5a-afc6-04e3e3c675d3.h5ad,visium__psc-PSC011__blockD1__4992sx35477g__721bfc1c-f77f-4c5a-afc6-04e3e3c675d3.h5ad,44.6,721bfc1c-f77f-4c5a-afc6-04e3e3c675d3,721bfc1c,https://datasets.cellxgene.cziscience.com/721bfc1c-f77f-4c5a-afc6-04e3e3c675d3.h5ad
|
| 13 |
+
7bc39166-b664-4c92-922d-f9a8047768d2.h5ad,visium__psc-PSC011__blockA1__4992sx35477g__7bc39166-b664-4c92-922d-f9a8047768d2.h5ad,61.8,7bc39166-b664-4c92-922d-f9a8047768d2,7bc39166,https://datasets.cellxgene.cziscience.com/7bc39166-b664-4c92-922d-f9a8047768d2.h5ad
|
| 14 |
+
7d4d0da4-655e-438a-a2ec-b4371e2b80fc.h5ad,sc__psc-pbc-healthy__all-cells__89637cx32596g__7d4d0da4-655e-438a-a2ec-b4371e2b80fc.h5ad,1119.0,7d4d0da4-655e-438a-a2ec-b4371e2b80fc,7d4d0da4,https://datasets.cellxgene.cziscience.com/7d4d0da4-655e-438a-a2ec-b4371e2b80fc.h5ad
|
| 15 |
+
9bc7506b-7e73-4cea-bbb3-3603a016fbca.h5ad,sc__healthy__lymphoid__16665cx32596g__9bc7506b-7e73-4cea-bbb3-3603a016fbca.h5ad,127.7,9bc7506b-7e73-4cea-bbb3-3603a016fbca,9bc7506b,https://datasets.cellxgene.cziscience.com/9bc7506b-7e73-4cea-bbb3-3603a016fbca.h5ad
|
| 16 |
+
9c731db0-9f44-4f8c-8139-9b9af2bcc782.h5ad,visium__healthy-C73__blockA1__4992sx35477g__9c731db0-9f44-4f8c-8139-9b9af2bcc782.h5ad,1108.6,9c731db0-9f44-4f8c-8139-9b9af2bcc782,9c731db0,https://datasets.cellxgene.cziscience.com/9c731db0-9f44-4f8c-8139-9b9af2bcc782.h5ad
|
| 17 |
+
a95e1659-9a48-4c55-8062-621ee4df9160.h5ad,sc__healthy__endothelial__9422cx32596g__a95e1659-9a48-4c55-8062-621ee4df9160.h5ad,77.7,a95e1659-9a48-4c55-8062-621ee4df9160,a95e1659,https://datasets.cellxgene.cziscience.com/a95e1659-9a48-4c55-8062-621ee4df9160.h5ad
|
| 18 |
+
b2287ef1-eac3-49cc-93de-65df74e26a61.h5ad,visium__healthy-C73__blockC1__4992sx35477g__b2287ef1-eac3-49cc-93de-65df74e26a61.h5ad,1172.4,b2287ef1-eac3-49cc-93de-65df74e26a61,b2287ef1,https://datasets.cellxgene.cziscience.com/b2287ef1-eac3-49cc-93de-65df74e26a61.h5ad
|
| 19 |
+
GSM4648565_liver_raw_counts.h5ad,sc__healthy-nat__liver__13083cx33694g__GSM4648565.h5ad,120.5,GSM4648565,GSM4648565,https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSM4648565
|
| 20 |
+
GSE125449_Set1_matrix.mtx.gz,sc__hcc-iccA-mixed__ma2019-set1-12pts__5115cx20124g__GSE125449-set1.h5ad,21.4,GSE125449-set1,sc-10x,https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE125449
|
| 21 |
+
GSE125449_Set2_matrix.mtx.gz,sc__hcc-iccA-mixed__ma2019-set2-7pts__4831cx19572g__GSE125449-set2.h5ad,15.7,GSE125449-set2,sc-10x,https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE125449
|
| 22 |
+
GSE151530_matrix.mtx.gz,sc__hcc-iccA-treated__ma2021-46pts__56721cx18667g__GSE151530.h5ad,229.8,GSE151530,sc-10x,https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE151530
|
| 23 |
+
GSE140228_UMI_counts_Droplet.mtx.gz,sc__hcc-iccA-cd45__sharma2020-droplet-5pts__66187cx54574g__GSE140228-droplet.h5ad,192.3,GSE140228-droplet,sc-10x,https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE140228
|
| 24 |
+
GSE140228_read_counts_Smartseq2.csv.gz,sc__hcc-cd45-ss2__sharma2020-ss2-6pts__7074cx54574g__GSE140228-ss2.h5ad,90.0,GSE140228-ss2,sc-smartseq2,https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE140228
|
_view_only_list.md
ADDED
|
@@ -0,0 +1,71 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# Liver_References · View-Only List
|
| 2 |
+
|
| 3 |
+
**Date**: 2026-05-25
|
| 4 |
+
**目的**: 标记本目录下**已 ingest 但 X/X.raw 都不是 integer raw counts** 的 .h5ad 文件——这些文件可以做 UMAP / clustering / 看 marker 表达 / cell-type 注释验证，但**不能直接用于**：
|
| 5 |
+
|
| 6 |
+
- DESeq2-style 差异表达（需 raw integer counts）
|
| 7 |
+
- 重新 normalize / batch correction（已经被作者 scale 过）
|
| 8 |
+
- 整合到 scVI / Scanorama / Harmony 的 raw-counts pipeline（数值范围不对）
|
| 9 |
+
|
| 10 |
+
下游 census 整合时**这些文件应该被 skip**，或者从原始 GEO mtx 重 ingest 后替换。
|
| 11 |
+
|
| 12 |
+
---
|
| 13 |
+
|
| 14 |
+
## ⚠️ 当前 View-Only 文件清单
|
| 15 |
+
|
| 16 |
+
| 文件 | X 状态 | .raw.X 状态 | 真 raw counts 来源 |
|
| 17 |
+
|---|---|---|---|
|
| 18 |
+
| `sc__hcc-tumor-normal__sharma2020-58pts__73589cx2608g__GSE156625-HCC.h5ad` | scaled HVG (2,608 genes, neg values −2.1~10) | log-normalized full (19,852 genes, max 7.0) | `_staging/wave2_geo/GSE156625/HCCmatrix.mtx.gz` (562 MB) + HCCbarcodes + HCCgenes |
|
| 19 |
+
| `sc__hcc-fetal__sharma2020-62pts__109238cx2384g__GSE156625-HCCF.h5ad` | scaled HVG (2,384 genes, neg values −1.3~10) | log-normalized full (20,417 genes, max 6.3) | `_staging/wave2_geo/GSE156625/HCCFmatrix.mtx.gz` (687 MB) + HCCFbarcodes + HCCFgenes |
|
| 20 |
+
|
| 21 |
+
---
|
| 22 |
+
|
| 23 |
+
## 怎么"反悔" — 拿回真 raw counts
|
| 24 |
+
|
| 25 |
+
两个文件源自 GEO GSE156625 (Sharma 2020 Onco-fetal Cell paper)。raw 10x matrix 在 staging 里已经下好：
|
| 26 |
+
|
| 27 |
+
```bash
|
| 28 |
+
ls /Users/zhongjunfu/Desktop/项目/AIVIN/database_unified/_staging/wave2_geo/GSE156625/
|
| 29 |
+
# GSE156625_HCCbarcodes.tsv.gz 129M cell barcodes
|
| 30 |
+
# GSE156625_HCCgenes.tsv.gz 259K gene symbols
|
| 31 |
+
# GSE156625_HCCmatrix.mtx.gz 562M integer counts
|
| 32 |
+
# GSE156625_HCCFbarcodes.tsv.gz 138M
|
| 33 |
+
# GSE156625_HCCFgenes.tsv.gz 259K
|
| 34 |
+
# GSE156625_HCCFmatrix.mtx.gz 687M
|
| 35 |
+
```
|
| 36 |
+
|
| 37 |
+
重 ingest 用 `scripts/ingest_wave2_remaining.py` 的 `_read_10x_triplet()` 函数即可——已经有现成代码。重 ingest 之前先备份现 .h5ad（含 louvain / UMAP / cell type annotations）到 `derived_views/sharma2020_processed_view/` 保留作者的预处理产物。
|
| 38 |
+
|
| 39 |
+
---
|
| 40 |
+
|
| 41 |
+
## 防御性使用规则（团队约定）
|
| 42 |
+
|
| 43 |
+
任何下游脚本读取 Liver_References 时，应先检查 `uns.x_state`：
|
| 44 |
+
|
| 45 |
+
```python
|
| 46 |
+
import anndata as ad
|
| 47 |
+
a = ad.read_h5ad(fp, backed='r')
|
| 48 |
+
if a.uns.get("x_state") in ("scaled_HVG", "scaled"):
|
| 49 |
+
# 这是 view-only file
|
| 50 |
+
if "raw_state" in a.uns and a.uns["raw_state"].startswith("log_normalized"):
|
| 51 |
+
# 可以用 .raw 做 score_genes 但不能做 DE
|
| 52 |
+
...
|
| 53 |
+
else:
|
| 54 |
+
# 不应用作 DE / re-normalize 输入
|
| 55 |
+
raise ValueError(f"{fp.name} has scaled X, no raw counts — re-ingest from _staging/")
|
| 56 |
+
```
|
| 57 |
+
|
| 58 |
+
或者读取本目录的 `_view_only_list.md`（本文档）跑前过滤。
|
| 59 |
+
|
| 60 |
+
---
|
| 61 |
+
|
| 62 |
+
## 未来加入此清单的判定规则
|
| 63 |
+
|
| 64 |
+
新 ingest 的 .h5ad 应该自动跑 `scripts/verify_h5ad.py --report`，若发现：
|
| 65 |
+
|
| 66 |
+
- `.X` 含负值 → scaled，自动加入 view-only
|
| 67 |
+
- `.X` 全为整数 + sparsity ≥ 90% → raw counts，正常
|
| 68 |
+
- `.X` 全为浮点 [0, 10] + sparsity ≥ 80% → log-normalized，**borderline**（看具体场景）
|
| 69 |
+
- `.X` 列数远小于 `.raw.X` 列数 → HVG subset 而非全基因，view-only
|
| 70 |
+
|
| 71 |
+
详见 `scripts/verify_h5ad.py` 自动判定逻辑。
|
merged_views/MERGE_REPORT.md
ADDED
|
@@ -0,0 +1,190 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# MERGE_REPORT · Liver_References 同类项合并
|
| 2 |
+
|
| 3 |
+
**生成日期**：2026-05-15
|
| 4 |
+
**对象目录**：`database_unified/Liver_References/merged_views/`
|
| 5 |
+
**合并范围**：Liver_References 下"同性质"原文件的派生整合视图——三类合并对应 sc healthy（per-cell-type 子集统一）/ visium healthy（C73 三切片统一）/ visium PSC（PSC011 四切片统一）。
|
| 6 |
+
**硬约束**：**原 17 份文件未做任何修改**；本目录只放派生视图，与原文件平级独立存在。
|
| 7 |
+
|
| 8 |
+
---
|
| 9 |
+
|
| 10 |
+
## 一、合并范围与三个派生输出
|
| 11 |
+
|
| 12 |
+
| # | 合并类型 | 输入（原文件数）| 输出文件 | 输出体量 | 输出 cells/spots |
|
| 13 |
+
|---|---|---:|---|---:|---:|
|
| 14 |
+
| 1 | sc healthy per-cell-type 统一 | 8 | `sc__healthy__merged-all-celltypes__107542cx32596g.h5ad` | 319.3 MB | 107,542 cells × 32,596 genes |
|
| 15 |
+
| 2 | visium healthy C73 三切片统一 | 3 | `visium__healthy-C73__merged-all-blocks__14976sx35477g.h5ad` | 57.1 MB | 14,976 spots × 35,477 genes |
|
| 16 |
+
| 3 | visium PSC PSC011 四切片统一 | 4 | `visium__psc-PSC011__merged-all-blocks__19968sx35477g.h5ad` | 90.5 MB | 19,968 spots × 35,477 genes |
|
| 17 |
+
|
| 18 |
+
**总计**：15 个原文件参与了 3 次合并（剩下 2 个 all-cells 文件未参与合并——它们各自就是独立的全集，没有"同类项"可合）。所有 3 个派生输出共 466.9 MB，相对于原 6.96 GB 总体积是约 **6.7%** 的派生层占比。
|
| 19 |
+
|
| 20 |
+
---
|
| 21 |
+
|
| 22 |
+
## 二、为什么是这三个合并组、为什么有 2 个文件不进合并
|
| 23 |
+
|
| 24 |
+
### 进合并的 3 组同类项
|
| 25 |
+
|
| 26 |
+
**组 1 · sc healthy per-cell-type** 是 Liver_References 子包 A 的全部 8 份——B 细胞 / 胆管细胞 / 内皮 / 肝细胞 v1 / 肝细胞 v2 / 淋巴样 / 巨噬 / 肝星状。它们的共同点是 `modality=sc` + `cohort=healthy` + 上游已分 cell type 切片好。把这 8 份合并的物理动机是**做一份"健康肝脏完整 sc 参考"**——下游做"任意细胞类型在 BOVIN 82 节点空间里的分布"或者"健康肝 vs 病肝细胞类型组成对比"时，单一文件比 8 文件方便得多。
|
| 27 |
+
|
| 28 |
+
**组 2 · visium healthy** 是 C73 健康捐献者的 3 个连续切片（A1 / C1 / D1）。它们都是同一名供体在同一时段的不同物理切片，**做合并是为了把"C73 这个人的肝组织空间表达"作为一个整体处理**，下游做"healthy liver lobule 的标志区域"或"C73 内部空间一致性"分析时无需手动循环 3 文件。
|
| 29 |
+
|
| 30 |
+
**组 3 · visium PSC** 是 PSC011 患者的 4 个连续切片（A1 / B1 / C1 / D1）。逻辑同组 2——这是 PSC011 这一名患者的"完整 Visium 病例"。
|
| 31 |
+
|
| 32 |
+
### 不进合并的 2 个文件
|
| 33 |
+
|
| 34 |
+
**子包 B 的 2 个 all-cells 文件**（`sc__psc-pbc-healthy__all-cells__89637c...` 与 `sn__psc-pbc-healthy__all-cells__105780c...`）**各自已是同类全集**——它们已经在上游 paper 里跨 PSC + PBC + healthy 全部供体做完了集成，**没有"同类项"可合**。强行把它们与本批合并产物再合并不仅没有信息增益，反而会引入 sc-vs-sn 模态混淆。**留作独立的全集参考**，自己单独使用。
|
| 35 |
+
|
| 36 |
+
如果将来想做"sc 全集（B-1）+ sc healthy per-cell-type（合并组 1）+ sn 全集（B-2）"的跨模态统一图谱，那是 v2 合并任务，需要 batch correction（如 Harmony / scVI）介入，不在本次"同类项合并"的 scope 里。
|
| 37 |
+
|
| 38 |
+
---
|
| 39 |
+
|
| 40 |
+
## 三、每个合并都做了什么 + 没做什么（透明记录）
|
| 41 |
+
|
| 42 |
+
### 合并组 1 · sc healthy per-cell-type
|
| 43 |
+
|
| 44 |
+
**做了的事**：
|
| 45 |
+
|
| 46 |
+
读入 8 个原文件的 `X`（归一化表达）+ `obs`（细胞元信息）+ `var`（基因列表，8 份完全一致）。对每个细胞**加 3 列 obs provenance**——`source_subset`（如 `stellate` / `hepatocyte-v1`）、`source_uuid8`（原文件 UUID 前 8 位）、`cell_barcode_original`（原 cell barcode 备份）。细胞 index **加前缀 `<subset>|` 防止 barcode 重复**（不同 subset 可能用同一 barcode）。用 `anndata.concat(axis=0, join='inner', merge='unique')` 在基因维上做交集合并（实际上 8 份 var 完全一致，inner = outer）。`uns['merge_provenance']` 记录合并来源、日期、原因、注意事项。
|
| 47 |
+
|
| 48 |
+
**没做的事**：
|
| 49 |
+
|
| 50 |
+
第一，**没有把 raw counts（`raw.X`）打包进合并文件**——原 8 文件各自的 raw counts 仍在原文件 `raw.X` 里。这个取舍是为了控制合并文件体积 + 合并耗时（原计划包含但耗时超 45s）。需要 raw counts 做自定义归一化的用户，请回到 `Liver_References/` 的原文件按 source_uuid8 反查。
|
| 51 |
+
|
| 52 |
+
第二，**没有保留 per-celltype 的 `obsm` 嵌入**（X_pca / X_harmony / X_umap / X_varimax）——这些嵌入是在**每个 cell type 子集内部**计算的（即"只在肝细胞群内做的 UMAP"vs"只在巨噬群内做的 UMAP"），跨 cell type 没有可比性。合并后保留它们反而会误导分析。下游用户应在合并对象上**重新跑 PCA + Harmony + UMAP**（标准 scanpy 流程，约 5–10 分钟在 32 GB RAM 机器上）。
|
| 53 |
+
|
| 54 |
+
第三，**没有做 batch correction**——8 个原文件用了不同的 10x 化学（3' v2 / 3' v3 / 5' v1）。合并对象的 `X` 直接是原归一化值串接，跨 subset 比较时存在化学批次效应。要做精细的跨 subset 分析建议跑 Harmony 用 `assay` 字段作 batch key。
|
| 55 |
+
|
| 56 |
+
### 合并组 2 · visium healthy C73 三切片
|
| 57 |
+
|
| 58 |
+
**做了的事**：
|
| 59 |
+
|
| 60 |
+
用 `h5py` 直接读 `X` / `obs` / `var` / `obsm`（绕过 anndata `read_h5ad` 自动加载 uns 的开销——原文件 uns 含 2.6 GB fullres 图像）。每个 spot 加 3 列 obs provenance——`block_id`（`blockA1` / `blockC1` / `blockD1`）、`source_uuid8`、`cell_barcode_original`。spot index 加前缀 `<block_id>|` 防重复。3 文件 var 完全一致（35,477 基因），inner join 直接 concat。`uns['spatial']` 用**字典形式**保留三个 library 的 hires 缩略图 + scalefactors——这是 scanpy/squidpy 处理 multi-library Visium 的标准格式。
|
| 61 |
+
|
| 62 |
+
**没做的事**：
|
| 63 |
+
|
| 64 |
+
第一，**fullres 图像（每切片 2.6 GB，3 切片共 ~8 GB）从合并文件里 drop 掉了**——`uns['spatial'][<lib_id>]['images']` 只保留 `hires` 缩略图（每切片 ~12 MB，3 切片 ~36 MB）。原文件 fullres 完整保留。如果下游分析需要 fullres（如做高分辨率细胞分割），请按 `source_uuid8` 回原文件取。
|
| 65 |
+
|
| 66 |
+
第二，**没有做 Visium spot 之间的跨切片对齐**——3 个切片在物理上是连续的（A1/C1/D1 是同一组织块上不同位置），但它们的 `obsm['spatial']` 坐标是各切片独立的 pixel 坐标系，跨切片直接比较坐标无意义。下游想做"跨切片 3D 对齐"需要专门的 spatial registration 算法（如 PASTE / STalign），不在合并 scope 里。
|
| 67 |
+
|
| 68 |
+
第三，**没有做 spot 级别的 H&E 图像融合**——3 切片各自有独立的 hires 病理图，合并对象只是把它们字典存放，没有拼接。视化时可以并排展示，但物理对齐需要后处理。
|
| 69 |
+
|
| 70 |
+
### 合并组 3 · visium PSC PSC011 四切片
|
| 71 |
+
|
| 72 |
+
技术处理与合并组 2 完全相同。唯一差异：4 个 PSC011 原文件**不含 fullres 图像**（只有 hires），所以合并时没有 fullres 可以 drop。合并文件保留全部 4 个 library 的 hires 图像 + scalefactors。
|
| 73 |
+
|
| 74 |
+
---
|
| 75 |
+
|
| 76 |
+
## 四、合并文件的 obs schema（统一约定）
|
| 77 |
+
|
| 78 |
+
合并后的所有 3 个文件遵循统一的 provenance schema：
|
| 79 |
+
|
| 80 |
+
| 列名 | 含义 | 出现在哪个合并文件 |
|
| 81 |
+
|---|---|---|
|
| 82 |
+
| `source_subset` | 该细胞 / spot 原属哪个 sub-type（如 `stellate`, `hepatocyte-v1`） | sc healthy |
|
| 83 |
+
| `block_id` | 该 spot 原属哪个切片（`blockA1` / `blockB1` / `blockC1` / `blockD1`） | visium healthy + PSC |
|
| 84 |
+
| `source_uuid8` | 该细胞 / spot 原文件 UUID 前 8 位（用于反查 `_provenance/uuid_to_new_name.csv`） | 所有 3 个合并文件 |
|
| 85 |
+
| `cell_barcode_original` | 原始 cell barcode（合并后被前缀化以保证唯一性，此列保留原值） | 所有 3 个合并文件 |
|
| 86 |
+
| 其余 obs 列 | 直接继承自原文件（disease / tissue / cell_type / donor_id 等） | 所有 3 个合并文件 |
|
| 87 |
+
|
| 88 |
+
合并文件的 cell/spot index 格式：`<source_subset或block_id>|<original_barcode>`，全局唯一。
|
| 89 |
+
|
| 90 |
+
---
|
| 91 |
+
|
| 92 |
+
## 五、对账与完整性核验
|
| 93 |
+
|
| 94 |
+
### 数量对账
|
| 95 |
+
|
| 96 |
+
```
|
| 97 |
+
合并组 1 · sc healthy:
|
| 98 |
+
b-cell: 1,250
|
| 99 |
+
cholangiocyte: 1,011
|
| 100 |
+
endothelial: 9,422
|
| 101 |
+
hepatocyte-v1: 53,015
|
| 102 |
+
hepatocyte-v2: 13,635
|
| 103 |
+
lymphoid: 16,665
|
| 104 |
+
macrophage: 11,127
|
| 105 |
+
stellate: 1,417
|
| 106 |
+
-----------------------
|
| 107 |
+
合并文件总细胞数: 107,542 ✅ 与 8 文件细胞数之和一致
|
| 108 |
+
|
| 109 |
+
合并组 2 · visium healthy:
|
| 110 |
+
blockA1: 4,992
|
| 111 |
+
blockC1: 4,992
|
| 112 |
+
blockD1: 4,992
|
| 113 |
+
-----------------------
|
| 114 |
+
合并文件总 spot 数: 14,976 ✅ 与 3 文件 spot 数之和一致
|
| 115 |
+
|
| 116 |
+
合并组 3 · visium PSC:
|
| 117 |
+
blockA1: 4,992
|
| 118 |
+
blockB1: 4,992
|
| 119 |
+
blockC1: 4,992
|
| 120 |
+
blockD1: 4,992
|
| 121 |
+
-----------------------
|
| 122 |
+
合并文件总 spot 数: 19,968 ✅ 与 4 文件 spot 数之和一致
|
| 123 |
+
```
|
| 124 |
+
|
| 125 |
+
### 原文件完整性
|
| 126 |
+
|
| 127 |
+
合并完成后，`Liver_References/` 下 17 个原 `.h5ad` 文件**逐一核对未被修改**（mtime 未变、size_mb 未变）。如有怀疑可对照 `_provenance/uuid_to_new_name.csv` 的 `size_mb` 列做完整性核对。
|
| 128 |
+
|
| 129 |
+
### 基因列表一致性
|
| 130 |
+
|
| 131 |
+
- sc healthy 合并：8 文件 var 完全一致（32,596 基因 Ensembl + HGNC），合并后 var = 32,596 基因 ✅
|
| 132 |
+
- visium healthy 合并：3 文件 var 完全一致（35,477 基因），合并后 var = 35,477 基因 ✅
|
| 133 |
+
- visium PSC 合并：4 文件 var 完全一致（35,477 基因），合并后 var = 35,477 基因 ✅
|
| 134 |
+
|
| 135 |
+
跨合并组（sc 32,596 vs visium 35,477）的合并不在本次 scope；如未来要做，需先对 Ensembl ID 取交集。
|
| 136 |
+
|
| 137 |
+
---
|
| 138 |
+
|
| 139 |
+
## 六、合并文件之间的关系图
|
| 140 |
+
|
| 141 |
+
```
|
| 142 |
+
Liver_References/
|
| 143 |
+
├── (17 原文件，不变)
|
| 144 |
+
│
|
| 145 |
+
└── merged_views/ ← 本次新增派生层
|
| 146 |
+
├── MERGE_REPORT.md ← 本文档
|
| 147 |
+
│
|
| 148 |
+
├── sc__healthy__merged-all-celltypes__107542cx32596g.h5ad
|
| 149 |
+
│ ↑ 来自 8 个 sc__healthy__*.h5ad（cell type subsets）
|
| 150 |
+
│
|
| 151 |
+
├── visium__healthy-C73__merged-all-blocks__14976sx35477g.h5ad
|
| 152 |
+
│ ↑ 来自 3 个 visium__healthy-C73__*.h5ad（C73 切片）
|
| 153 |
+
│
|
| 154 |
+
└── visium__psc-PSC011__merged-all-blocks__19968sx35477g.h5ad
|
| 155 |
+
↑ 来自 4 个 visium__psc-PSC011__*.h5ad（PSC011 切片）
|
| 156 |
+
|
| 157 |
+
未参与本次合并的 2 个原文件：
|
| 158 |
+
sc__psc-pbc-healthy__all-cells__89637cx32596g__... ← 已是 sc 全集
|
| 159 |
+
sn__psc-pbc-healthy__all-cells__105780cx32596g__... ← 已是 sn 全集
|
| 160 |
+
```
|
| 161 |
+
|
| 162 |
+
---
|
| 163 |
+
|
| 164 |
+
## 七、可能的下一步合并（v2 提议，不在本次 scope）
|
| 165 |
+
|
| 166 |
+
如果未来要把"派生视图"做得更深，可以考虑以下三类**跨子包合并**——但每一项都需要更复杂的处理（batch correction / 模态对齐 / 基因交集）：
|
| 167 |
+
|
| 168 |
+
**v2 候选 #1 · sc + sn 全集联合**——把 `sc__psc-pbc-healthy__all-cells` 与 `sn__psc-pbc-healthy__all-cells` 用 Harmony / scVI 做跨模态 latent space，产出"PSC+PBC+healthy 全细胞 + 全核统一图谱"。难点：sc 与 sn 的基因检出模式系统差异。
|
| 169 |
+
|
| 170 |
+
**v2 候选 #2 · 健康 vs PSC visium 跨疾病合并**——把 `visium__healthy-C73` 与 `visium__psc-PSC011` 合并成"健康 vs PSC 空间对比图谱"。难点：跨患者切片对齐 + spot 级别 batch effect。
|
| 171 |
+
|
| 172 |
+
**v2 候选 #3 · 健康肝完整图谱**——sc healthy 合并视图（本组 1）+ sc all-cells 中的健康部分（24,241 cells）+ sn all-cells 中的健康部分（26,515 nuclei）= 完整健康肝多模态参考。难点：sc/sn 跨模态 + 跨 chemistry batch。
|
| 173 |
+
|
| 174 |
+
以上 v2 工作建议另起 session 单独处理，每项至少 30–60 分钟。
|
| 175 |
+
|
| 176 |
+
---
|
| 177 |
+
|
| 178 |
+
## 八、对应的可视化资产
|
| 179 |
+
|
| 180 |
+
本次合并同时生成了一份 HTML dashboard：
|
| 181 |
+
|
| 182 |
+
```
|
| 183 |
+
database_unified/Liver_References/merged_views/dashboard.html
|
| 184 |
+
```
|
| 185 |
+
|
| 186 |
+
含合并对照表、cell type 组成饼图、供体分布柱图、7 张 Visium 切片 hires 缩略图。**单文件自包含**，浏览器直接打开（无需任何后端）。
|
| 187 |
+
|
| 188 |
+
---
|
| 189 |
+
|
| 190 |
+
*合并工作执行于 2026-05-15 · 共 3 次合并 · 输入 15 文件 / 输出 3 派生文件 / 派生体积 467 MB · 原 17 文件未做任何修改 · 完整 provenance 记录在每份合并文件 uns["merge_provenance"]。*
|
merged_views/dashboard.html
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
preview.png
ADDED
|
Git LFS Details
|
sc__hcc-cd45-ss2__sharma2020-ss2-6pts__7074cx54574g__GSE140228-ss2.h5ad
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:52b8453dea08caf1d708f770cc37f0a620446d439eb956a17a913be567fc1799
|
| 3 |
+
size 94321756
|
sc__hcc-cd45__guo2025-9pts__191435cx28671g__GSE235863-9pt.h5ad
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:95e32c51b088a4c958c26a7806880e3b37221570ce605acc13aeef5e686a7b98
|
| 3 |
+
size 771001429
|
sc__hcc-cd8tcell__guo2025-5pts__95408cx26256g__GSE235863-5pt.h5ad
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2565491bb97d8e37b0480072fd8d6e3338e85e3b8c62795ff308c1e58287783f
|
| 3 |
+
size 336333290
|
sc__hcc-fetal__sharma2020-62pts__109238cx2384g__GSE156625-HCCF.h5ad
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9d3d45363d6477cdac3338d4dd27e2956b6ac8639cd9acfc7cb0a1d635f2159f
|
| 3 |
+
size 1183864996
|
sc__hcc-hbv-caf__zhang2023-11pts__115732cx36601g__GSE202642.h5ad
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:740be1dbe5e83b11505d672bc00c3f164f0c2e2c5bf9397ad8901b8ab82d87cc
|
| 3 |
+
size 366462969
|
sc__hcc-iccA-mixed__ma2019-set1-12pts__5115cx20124g__GSE125449-set1.h5ad
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c7acbd3afe684859141d5d78e13315f43f8359d111a751375b04190244632b95
|
| 3 |
+
size 22423207
|
sc__hcc-iccA-treated__ma2021-46pts__56721cx18667g__GSE151530.h5ad
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ac13a226ab33a77eeb9af14b34b317c24b949a9fabc2a19c29ccfe3c1ce8a229
|
| 3 |
+
size 240916612
|
sc__hcc-mash-spectrum__huang2025-4pts__34396cx32285g__GSE282630.h5ad
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2d0ff647ab7ef32bac00c1ac2928609eac66c10e61cb707ee7d2874b96686da7
|
| 3 |
+
size 174313350
|
sc__hcc-multisite__lu2022-10pts__71915cx25712g__GSE149614.h5ad
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:39d304aa7dde7eb663778ce714390993a37bfc6892e9a452d75239bf908931ae
|
| 3 |
+
size 344066570
|
sc__hcc-tcell__zheng2017-6pts__5063cx23459g__GSE98638.h5ad
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:881147506928ddce17ef1df0f78ffe1ce4b767610c884012e2b4fcbca52806d8
|
| 3 |
+
size 58693651
|
sc__hcc-trm__park2025-7pts__41848cx36601g__GSE281110.h5ad
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9289047df8f17703adc6beadc924d6399a3445288d2631a39b21f76e2639783d
|
| 3 |
+
size 164385165
|
sc__healthy-nat__liver__13083cx33694g__GSM4648565.h5ad
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:58d1bf58c5d37c4515b512416aafd752325897e627b523344f494ae748ffff6e
|
| 3 |
+
size 126649408
|
sc__healthy__b-cell__1250cx32596g__6ade4ff5-368c-4276-b051-818dc954da6d.h5ad
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:aff78ed1b1b1bfdbf944142589431d685ef87a6b868727a9570c53b93c5ec415
|
| 3 |
+
size 18101688
|
sc__healthy__cholangiocyte__1011cx32596g__601ef580-74ce-4a87-96e8-8e22bf4ed9fa.h5ad
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c83d3f3aa25b6062d9b63a2901c32fa93863a7f3cd45eeeebfd604cac34a4468
|
| 3 |
+
size 16684975
|
sc__healthy__endothelial__9422cx32596g__a95e1659-9a48-4c55-8062-621ee4df9160.h5ad
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c07dc58bd96f2668ddb344e5d8ca810306f670f63a500c1089f1bddd28f1224c
|
| 3 |
+
size 77659987
|
sc__healthy__hepatocyte-v1__53015cx32596g__63137dcf-6236-464d-9018-e58c9323f59c.h5ad
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f92e71176d98480a4aeb3229acc7632ce44bfee6884943a09c96e07b848b5ac6
|
| 3 |
+
size 439781839
|
sc__healthy__hepatocyte-v2__13635cx32596g__6b241dde-25c0-4edc-9021-03603ec3a524.h5ad
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6a19c26bb9fc37baa3e90ad81f3aa9552b283548875e468fbf1d107aeb03845f
|
| 3 |
+
size 85095660
|
sc__healthy__lymphoid__16665cx32596g__9bc7506b-7e73-4cea-bbb3-3603a016fbca.h5ad
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2a3c483b6fb51bd1fafd511b5b4782b44609d480a2ddf0aca44ed7e0f7051a85
|
| 3 |
+
size 127744689
|
sc__healthy__macrophage__11127cx32596g__2cfec927-9163-4684-ae04-c15175a6d781.h5ad
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f5ac19564f638d167de420bd9a79288a1bb7475866585b0a6df1661c81c3be12
|
| 3 |
+
size 93688946
|
sc__healthy__stellate__1417cx32596g__08a9f031-1e76-405a-8050-0635743ce187.h5ad
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:368636e54d162ab5f4735f6451f3bc471043508d80110877056091b311af073c
|
| 3 |
+
size 14693734
|
sc__psc-pbc-healthy__all-cells__89637cx32596g__7d4d0da4-655e-438a-a2ec-b4371e2b80fc.h5ad
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:af281ee9579086e671d9b07c1fed487517212721f221824de8e8f1c632d3c6e0
|
| 3 |
+
size 1118971912
|
sn__hcc-tumor-normal__alvarez2022-1pts__39995cx58100g__GSE189175.h5ad
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:43f4984dcdcf58ff0c5ce04fe4ea388ac729249f1bd2f25950f8c1ae8ae88cb9
|
| 3 |
+
size 121474423
|
sn__hcc-tumor-normal__alvarez2022-3pts__39995cx58100g__GSE189175.h5ad
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a4d8fe053069b6604d601c736fdd0d706c1830364b2914af5a21985ba2415fef
|
| 3 |
+
size 121484412
|
sn__psc-pbc-healthy__all-cells__105780cx32596g__4b5895d7-6d92-471a-b13a-5c59a000ddc4.h5ad
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:87b325456d037c4a4463f36d91942cae4d6d4748b8cb5afa2e4c4134d4420199
|
| 3 |
+
size 1543814986
|
visium__hcc-antiPD1__liu2025-1pts__1320cx36601g__GSE238264-HCC5NR.h5ad
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:055fc4a0465abc269e7d2312093680e873c50f9582425762b2dc8fea0cd7e2ce
|
| 3 |
+
size 16005069
|
visium__hcc-antiPD1__liu2025-1pts__2170cx36601g__GSE238264-HCC3R.h5ad
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:830f862c0d60eed55d34fd4203ed9ebe148a08ae1d885de7f1e325aec9b4e2c3
|
| 3 |
+
size 19271492
|
visium__hcc-antiPD1__liu2025-1pts__2453cx36601g__GSE238264-HCC7NR.h5ad
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:37acac3996f904ac222a0869feac4a205c9ec0e036b56155979687eef39c5c77
|
| 3 |
+
size 25043672
|
visium__hcc-antiPD1__liu2025-1pts__2575cx36601g__GSE238264-HCC6NR.h5ad
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2412d315e218fcc230e49a36f615686e0e1b68a2cc9f65edba20f08dfa7cbd38
|
| 3 |
+
size 38629380
|
visium__hcc-antiPD1__liu2025-1pts__2766cx36601g__GSE238264-HCC2R.h5ad
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d7048ad790364dc75df2b33951a6db39bf1a9eac134c37a0471b0a224c822239
|
| 3 |
+
size 15889732
|
visium__hcc-antiPD1__liu2025-1pts__3002cx36601g__GSE238264-HCC4R.h5ad
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a5f0faa224184d14f9fdda57615defc7145f47717615e8e12da1c56a5ade622d
|
| 3 |
+
size 42219913
|
visium__hcc-antiPD1__liu2025-1pts__3006cx36601g__GSE238264-HCC1R.h5ad
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5ea5f226abfef74d496f2e5186d6840e6b4e71bebc985c5ee6c3f75902309790
|
| 3 |
+
size 23707846
|
visium__healthy-C73__blockA1__4992sx35477g__9c731db0-9f44-4f8c-8139-9b9af2bcc782.h5ad
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4d050ce4434f8c30034ea9ba44ffc6a3158f7ce2d88e65362d5227660f032efa
|
| 3 |
+
size 1108606004
|
visium__healthy-C73__blockC1__4992sx35477g__b2287ef1-eac3-49cc-93de-65df74e26a61.h5ad
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:570d4c05951583fdb98ebc63472164de7d3b7a03f7df6efb91b658ac392fbef7
|
| 3 |
+
size 1172387643
|
visium__healthy-C73__blockD1__4992sx35477g__44453de7-1d66-4bd1-a83d-0b73a1690d57.h5ad
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c4089fae3395aef106f9a767d370f1509a53e0cfab7a6f028cb319e606b3a7e4
|
| 3 |
+
size 1096878083
|
visium__psc-PSC011__blockA1__4992sx35477g__7bc39166-b664-4c92-922d-f9a8047768d2.h5ad
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:51856f0c82905e2cd0dd1e5fab9ecf58b7d055ab81b85c0329c263413a036ec9
|
| 3 |
+
size 61834920
|
visium__psc-PSC011__blockB1__4992sx35477g__6da751d4-63af-43b8-96db-395ca73dfb5f.h5ad
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:84214dd34dc0db7fba68462cfe1bf85c6bcd85dc28cf57d53d0727dcef45028e
|
| 3 |
+
size 52301622
|
visium__psc-PSC011__blockC1__4992sx35477g__0cc59004-2b35-4767-8278-83e097ef32d1.h5ad
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ad23a4da3bab4c159068a1c24b00bdacec6e34c091a1b68067401f50899e81a5
|
| 3 |
+
size 53259985
|