zzq1zh commited on
Commit
61f4d6c
·
verified ·
1 Parent(s): 71bb5de

Upload updated eccDNA dataset

Browse files
README.md CHANGED
@@ -1,3 +1,76 @@
1
- ---
2
- license: apache-2.0
3
- ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: apache-2.0
3
+ tags:
4
+ - biology
5
+ - genomics
6
+ - dna
7
+ - eccdna
8
+ size_categories:
9
+ - 10K<n<1M
10
+ task_categories:
11
+ - token-classification
12
+ ---
13
+
14
+ # Real vs. Pseudo-eccDNA Discrimination (Homo sapiens)
15
+
16
+ This dataset supports the **Real vs. Pseudo-eccDNA Discrimination** task for human eccDNA.
17
+ The goal is to train models that can distinguish true eccDNA sequences from pseudo-eccDNAs
18
+ randomly extracted from linear genomic regions with matched length distributions.
19
+
20
+ Each entry contains:
21
+ - `sequence`: raw eccDNA sequence (A/T/C/G)
22
+ - `label`:
23
+ - `1` → Real eccDNA
24
+ - `0` → Pseudo-eccDNA (negative control)
25
+
26
+ ---
27
+
28
+ ## 📁 Folder Structure
29
+ <pre>
30
+ real_vs_pseudo_eccdna_discrimination_human/
31
+ ├── data/
32
+ │ └── real_vs_pseudo_eccdna_discrimination_human.csv
33
+ └── README.md
34
+ </pre>
35
+
36
+ ---
37
+
38
+ ## 🚀 Quick Usage
39
+ <pre><code class="language-python">
40
+ from datasets import load_dataset, load_from_disk
41
+
42
+ # Load from local
43
+ dataset = load_from_disk("real_vs_pseudo_eccdna_discrimination_human/hf_data")
44
+
45
+ # Load from Hugging Face Hub (after upload)
46
+ dataset = load_dataset("your-username/real_vs_pseudo_eccdna_discrimination_human")
47
+
48
+ # Example: view label distribution
49
+ df = dataset["train"].to_pandas()
50
+ print(df['label'].value_counts())
51
+ </code></pre>
52
+
53
+ ---
54
+
55
+ ## Task Description
56
+
57
+ True eccDNAs are experimentally verified circular DNA molecules,
58
+ whereas pseudo-eccDNAs are generated by randomly extracting linear genomic segments
59
+ to match the true eccDNA length distribution.
60
+ This task assesses a model’s ability to capture **circular topology** and **regulatory context**
61
+ beyond simple sequence composition.
62
+
63
+ ---
64
+
65
+ ## Citation
66
+ If you use this dataset, please cite:
67
+ <pre><code class="language-python">
68
+
69
+ @inproceedings{liu2025eccdnamamba,
70
+ title={eccDNAMamba: A Pre-Trained Model for Ultra-Long eccDNA Sequence Analysis},
71
+ author={Zhenke Liu and Jien Li and Ziqi Zhang},
72
+ booktitle={ICML 2025 GenBio Workshop},
73
+ year={2025},
74
+ url={https://openreview.net/forum?id=56xKN7KJjy}
75
+ }
76
+ </code></pre>
data/real_vs_pseudo_eccdna_homo_sapiens.csv ADDED
The diff for this file is too large to render. See raw diff