Create scorer.py

scorer.py

@@ -0,0 +1,248 @@
+from __future__ import annotations
+
+import json
+from dataclasses import dataclass
+from typing import Dict, List
+
+import pandas as pd
+
+
+REQUIRED_COLS = [
+    "row_id",
+    "series_id",
+    "timepoint_h",
+    "organism",
+    "strain_id",
+    "drug_name",
+    "target_gene",
+    "mutation_id",
+    "mutant_frac_target_mod",
+    "mic_drug_mg_L",
+    "resistant_mic_cutoff_mg_L",
+    "media",
+    "assay_method",
+    "source_type",
+    "penetration_signal",
+    "earliest_penetration",
+]
+
+
+@dataclass
+class Thresholds:
+    min_points: int = 3
+
+    frac_threshold: float = 0.20
+    mic_fold_max_at_penetration: float = 2.0  # relative to baseline
+
+    confirm_requires_resistant: bool = True
+
+    spike_frac_min: float = 0.80
+    snapback_frac_max: float = 0.05
+
+
+def _validate(df: pd.DataFrame) -> List[str]:
+    errs: List[str] = []
+    missing = [c for c in REQUIRED_COLS if c not in df.columns]
+    if missing:
+        errs.append(f"missing_columns: {missing}")
+
+    for c in ["timepoint_h", "mutant_frac_target_mod", "mic_drug_mg_L", "resistant_mic_cutoff_mg_L"]:
+        if c in df.columns and df[c].isna().any():
+            errs.append(f"null_values_in: {c}")
+
+    if "mutant_frac_target_mod" in df.columns:
+        bad = ((df["mutant_frac_target_mod"] < 0) | (df["mutant_frac_target_mod"] > 1)).sum()
+        if bad:
+            errs.append(f"mutant_frac_out_of_range count={int(bad)}")
+
+    for c in ["mic_drug_mg_L", "resistant_mic_cutoff_mg_L"]:
+        if c in df.columns:
+            bad = (df[c] <= 0).sum()
+            if bad:
+                errs.append(f"non_positive_values_in: {c} count={int(bad)}")
+
+    for c in ["penetration_signal", "earliest_penetration"]:
+        if c in df.columns:
+            bad = (~df[c].isin([0, 1])).sum()
+            if bad:
+                errs.append(f"non_binary_values_in: {c} count={int(bad)}")
+
+    counts = df.groupby("series_id")["earliest_penetration"].sum()
+    bad_series = counts[counts > 1].index.tolist()
+    if bad_series:
+        errs.append(f"multiple_earliest_penetration_in_series: {bad_series}")
+
+    return errs
+
+
+def _flag_spike_snap(g: pd.DataFrame, t: Thresholds) -> pd.Series:
+    flag = pd.Series([0] * len(g), index=g.index)
+    if len(g) < 3:
+        return flag
+
+    g = g.sort_values("timepoint_h").copy()
+    for i in range(1, len(g) - 1):
+        idx = g.index[i]
+        prev_idx = g.index[i - 1]
+        next_idx = g.index[i + 1]
+
+        prev_v = float(g.loc[prev_idx, "mutant_frac_target_mod"])
+        v = float(g.loc[idx, "mutant_frac_target_mod"])
+        next_v = float(g.loc[next_idx, "mutant_frac_target_mod"])
+
+        spike = v >= t.spike_frac_min and prev_v <= t.snapback_frac_max and next_v <= t.snapback_frac_max
+        if spike:
+            flag.loc[idx] = 1
+
+    return flag
+
+
+def _f1(tp: int, fp: int, fn: int) -> float:
+    denom = 2 * tp + fp + fn
+    return 0.0 if denom == 0 else (2 * tp) / denom
+
+
+def score(path: str) -> Dict[str, object]:
+    df = pd.read_csv(path)
+    errors = _validate(df)
+    if errors:
+        return {"ok": False, "errors": errors}
+
+    t = Thresholds()
+
+    df = df.sort_values(["series_id", "timepoint_h"]).reset_index(drop=True)
+    df["pred_earliest_penetration"] = 0
+    df["pred_penetration_signal"] = 0
+    df["flag_measurement_spike"] = 0
+
+    series_rows: List[Dict[str, object]] = []
+
+    for sid, g in df.groupby("series_id"):
+        g = g.sort_values("timepoint_h").copy()
+        df.loc[g.index, "flag_measurement_spike"] = _flag_spike_snap(g, t).astype(int)
+
+        if len(g) < t.min_points:
+            series_rows.append(
+                {
+                    "series_id": sid,
+                    "y_pen": int(g["penetration_signal"].max()),
+                    "p_pen": 0,
+                    "true_transition_row_id": (str(g[g["earliest_penetration"] == 1].iloc[0]["row_id"]) if (g["earliest_penetration"] == 1).any() else None),
+                    "pred_transition_row_id": None,
+                    "flags": ["too_few_points"],
+                }
+            )
+            continue
+
+        base = g.iloc[0]
+        base_mic = float(base["mic_drug_mg_L"])
+        cutoff = float(base["resistant_mic_cutoff_mg_L"])
+
+        # if baseline already resistant, not an early penetration case
+        if base_mic >= cutoff:
+            series_rows.append(
+                {
+                    "series_id": sid,
+                    "y_pen": int(g["penetration_signal"].max()),
+                    "p_pen": 0,
+                    "true_transition_row_id": (str(g[g["earliest_penetration"] == 1].iloc[0]["row_id"]) if (g["earliest_penetration"] == 1).any() else None),
+                    "pred_transition_row_id": None,
+                    "flags": ["baseline_resistant"],
+                }
+            )
+            continue
+
+        hits = []
+        for idx, row in g.iterrows():
+            if idx == g.index[0]:
+                continue
+            if int(df.loc[idx, "flag_measurement_spike"]) == 1:
+                continue
+
+            frac = float(row["mutant_frac_target_mod"])
+            mic = float(row["mic_drug_mg_L"])
+
+            mic_fold = mic / base_mic if base_mic > 0 else 99.0
+
+            candidate = (
+                frac >= t.frac_threshold
+                and mic_fold <= t.mic_fold_max_at_penetration
+                and mic < cutoff
+            )
+
+            if candidate:
+                hits.append(idx)
+
+        confirm = True
+        if t.confirm_requires_resistant and hits:
+            later = g[g.index > hits[0]]
+            confirm = bool((later["mic_drug_mg_L"] >= cutoff).any())
+
+        if hits and confirm:
+            first = hits[0]
+            df.loc[first, "pred_earliest_penetration"] = 1
+            df.loc[g[g.index >= first].index, "pred_penetration_signal"] = 1
+
+        y = int(g["penetration_signal"].max())
+        p = int(df.loc[g.index, "pred_penetration_signal"].max())
+
+        true_tr = g[g["earliest_penetration"] == 1]
+        true_id = str(true_tr.iloc[0]["row_id"]) if len(true_tr) == 1 else None
+
+        pred_tr_rows = df.loc[g.index][df.loc[g.index, "pred_earliest_penetration"] == 1]
+        pred_id = str(pred_tr_rows.iloc[0]["row_id"]) if len(pred_tr_rows) == 1 else None
+
+        series_rows.append(
+            {
+                "series_id": sid,
+                "y_pen": y,
+                "p_pen": p,
+                "true_transition_row_id": true_id,
+                "pred_transition_row_id": pred_id,
+                "measurement_spike_flags": int(df.loc[g.index, "flag_measurement_spike"].sum()),
+            }
+        )
+
+    sr = pd.DataFrame(series_rows)
+
+    tp = int(((sr["y_pen"] == 1) & (sr["p_pen"] == 1)).sum())
+    fp = int(((sr["y_pen"] == 0) & (sr["p_pen"] == 1)).sum())
+    fn = int(((sr["y_pen"] == 1) & (sr["p_pen"] == 0)).sum())
+    tn = int(((sr["y_pen"] == 0) & (sr["p_pen"] == 0)).sum())
+
+    transition_hit = int(
+        (
+            sr["true_transition_row_id"].notna()
+            & (sr["true_transition_row_id"] == sr["pred_transition_row_id"])
+        ).sum()
+    )
+    transition_miss = int(
+        (
+            sr["true_transition_row_id"].notna()
+            & (sr["true_transition_row_id"] != sr["pred_transition_row_id"])
+        ).sum()
+    )
+
+    return {
+        "ok": True,
+        "path": path,
+        "counts": {"tp": tp, "fp": fp, "fn": fn, "tn": tn},
+        "metrics": {
+            "f1_series": _f1(tp, fp, fn),
+            "transition_hit": transition_hit,
+            "transition_miss": transition_miss,
+            "n_series": int(len(sr)),
+        },
+        "series_table": series_rows,
+    }
+
+
+if __name__ == "__main__":
+    import argparse
+
+    ap = argparse.ArgumentParser()
+    ap.add_argument("--path", required=True)
+    args = ap.parse_args()
+
+    result = score(args.path)
+    print(json.dumps(result, indent=2))