Create scorer.py

scorer.py

@@ -0,0 +1,252 @@
+from __future__ import annotations
+
+import json
+from dataclasses import dataclass
+from typing import Dict, List, Optional
+
+import pandas as pd
+
+
+REQUIRED_COLS = [
+    "row_id",
+    "series_id",
+    "timepoint_h",
+    "host_model",
+    "abx_drug",
+    "abx_dose_mg",
+    "abx_conc_ng_mL",
+    "expected_abx_conc_ng_mL",
+    "abx_conc_deviation",
+    "concomitant_drug",
+    "concomitant_exposure_index",
+    "expected_interaction_shift",
+    "interaction_shift_deviation",
+    "ddi_coherence_index",
+    "host_stress_index",
+    "later_adverse_ddi_flag",
+    "assay_method",
+    "source_type",
+    "ddi_emergence_signal",
+    "earliest_ddi_emergence",
+]
+
+
+@dataclass
+class Thresholds:
+    min_points: int = 3
+
+    stress_min: float = 0.80
+    concom_min: float = 0.80
+
+    coherence_max: float = 0.40
+    shift_dev_min: float = 40.0
+    require_two_consecutive: bool = True
+
+    spike_shift_dev_min: float = 80.0
+    snapback_shift_dev_max: float = 3.0
+
+
+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 ["concomitant_exposure_index", "ddi_coherence_index", "host_stress_index"]:
+        if c in df.columns:
+            bad = ((df[c] < 0) | (df[c] > 1)).sum()
+            if bad:
+                errs.append(f"{c}_out_of_range count={int(bad)}")
+
+    for c in ["abx_dose_mg", "abx_conc_ng_mL", "expected_abx_conc_ng_mL"]:
+        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 ["later_adverse_ddi_flag", "ddi_emergence_signal", "earliest_ddi_emergence"]:
+        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_ddi_emergence"].sum()
+    bad_series = counts[counts > 1].index.tolist()
+    if bad_series:
+        errs.append(f"multiple_earliest_ddi_emergence_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]
+        next_idx = g.index[i + 1]
+        v = float(g.loc[idx, "interaction_shift_deviation"])
+        next_v = float(g.loc[next_idx, "interaction_shift_deviation"])
+        if v >= t.spike_shift_dev_min and abs(next_v) <= t.snapback_shift_dev_max:
+            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_ddi_emergence"] = 0
+    df["pred_ddi_emergence_signal"] = 0
+    df["flag_shift_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_shift_spike"] = _flag_spike_snap(g, t).astype(int)
+
+        if len(g) < t.min_points:
+            series_rows.append(
+                {
+                    "series_id": sid,
+                    "y_ddi": int(g["ddi_emergence_signal"].max()),
+                    "p_ddi": 0,
+                    "true_transition_row_id": (str(g[g["earliest_ddi_emergence"] == 1].iloc[0]["row_id"]) if (g["earliest_ddi_emergence"] == 1).any() else None),
+                    "pred_transition_row_id": None,
+                    "flags": ["too_few_points"],
+                }
+            )
+            continue
+
+        has_concom = bool((g["concomitant_exposure_index"] > 0).any()) and bool((g["concomitant_drug"].astype(str).str.lower() != "none").any())
+        if not has_concom:
+            series_rows.append(
+                {
+                    "series_id": sid,
+                    "y_ddi": int(g["ddi_emergence_signal"].max()),
+                    "p_ddi": 0,
+                    "true_transition_row_id": (str(g[g["earliest_ddi_emergence"] == 1].iloc[0]["row_id"]) if (g["earliest_ddi_emergence"] == 1).any() else None),
+                    "pred_transition_row_id": None,
+                    "flags": ["no_concomitant"],
+                }
+            )
+            continue
+
+        hit: Optional[int] = None
+        for i in range(1, len(g)):
+            idx = g.index[i]
+            if int(df.loc[idx, "flag_shift_spike"]) == 1:
+                continue
+
+            if float(df.loc[idx, "host_stress_index"]) < t.stress_min:
+                continue
+            if float(df.loc[idx, "concomitant_exposure_index"]) < t.concom_min:
+                continue
+
+            coh = float(df.loc[idx, "ddi_coherence_index"])
+            sdev = float(df.loc[idx, "interaction_shift_deviation"])
+
+            if coh > t.coherence_max:
+                continue
+            if abs(sdev) < t.shift_dev_min:
+                continue
+
+            if t.require_two_consecutive:
+                if i + 1 >= len(g):
+                    continue
+                idx2 = g.index[i + 1]
+                if int(df.loc[idx2, "flag_shift_spike"]) == 1:
+                    continue
+                if float(df.loc[idx2, "host_stress_index"]) < t.stress_min:
+                    continue
+                if float(df.loc[idx2, "concomitant_exposure_index"]) < t.concom_min:
+                    continue
+                coh2 = float(df.loc[idx2, "ddi_coherence_index"])
+                sdev2 = float(df.loc[idx2, "interaction_shift_deviation"])
+                if coh2 > t.coherence_max or abs(sdev2) < t.shift_dev_min:
+                    continue
+
+            hit = idx
+            break
+
+        confirm = False
+        if hit is not None:
+            later = g[g.index > hit]
+            confirm = bool((later["later_adverse_ddi_flag"] == 1).any())
+
+        if hit is not None and confirm:
+            df.loc[hit, "pred_earliest_ddi_emergence"] = 1
+            df.loc[g[g.index >= hit].index, "pred_ddi_emergence_signal"] = 1
+
+        y = int(g["ddi_emergence_signal"].max())
+        p = int(df.loc[g.index, "pred_ddi_emergence_signal"].max())
+
+        true_tr = g[g["earliest_ddi_emergence"] == 1]
+        true_id: Optional[str] = None
+        if len(true_tr) == 1:
+            true_id = str(true_tr.iloc[0]["row_id"])
+
+        pred_tr_rows = df.loc[g.index][df.loc[g.index, "pred_earliest_ddi_emergence"] == 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_ddi": y,
+                "p_ddi": p,
+                "true_transition_row_id": true_id,
+                "pred_transition_row_id": pred_id,
+                "shift_spike_flags": int(df.loc[g.index, "flag_shift_spike"].sum()),
+            }
+        )
+
+    sr = pd.DataFrame(series_rows)
+
+    tp = int(((sr["y_ddi"] == 1) & (sr["p_ddi"] == 1)).sum())
+    fp = int(((sr["y_ddi"] == 0) & (sr["p_ddi"] == 1)).sum())
+    fn = int(((sr["y_ddi"] == 1) & (sr["p_ddi"] == 0)).sum())
+    tn = int(((sr["y_ddi"] == 0) & (sr["p_ddi"] == 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))