#!/usr/bin/env python3 # -*- coding: utf-8 -*- """ DMRepurpRank-Kor (디엠리퍼프랭크코어) ----------------------------------- 비당뇨 약물 → 당뇨 효능 재포지셔닝 후보 자동 통합 ranking + 한국어 IIT/grant proposal card 생성 standalone Python CLI. 7+ source 통합: - DrugBank / OpenTargets (target, mechanism) - UK Biobank / KoGES GWAS (Mendelian Randomization) - CMAP / L1000 (transcriptomic signature reversal) - FAERS / HIRA / KIDS-KD (claims / RWE disproportionality) - IMPC (KO mouse phenotype) - KCD-7/8 (E11.x T2DM code signal) - Viability (generic / patent / KFDA / HIRA coverage) ⚠️ 본 도구는 연구용·참고용 hypothesis 생성기입니다. 임상의사결정·환자 처방·실제 약물 재포지셔닝 승인 결정에 사용 금지. """ from __future__ import annotations import argparse import csv import json import math import os import sys from typing import Any # --------------------------------------------------------------------------- # Paths # --------------------------------------------------------------------------- HERE = os.path.dirname(os.path.abspath(__file__)) DATA_DIR = os.path.join(HERE, "data") DISCLAIMER = ( "본 도구는 연구용·참고용 hypothesis 생성기입니다. " "임상의사결정·환자 처방·실제 약물 재포지셔닝 승인 결정에 사용 금지." ) # --------------------------------------------------------------------------- # CSV loaders (stdlib csv — robust without pandas dependency) # --------------------------------------------------------------------------- def _load_csv(path: str) -> list[dict[str, str]]: with open(path, "r", encoding="utf-8") as f: reader = csv.DictReader(f) return list(reader) def load_all() -> dict[str, list[dict[str, str]]]: return { "drugs": _load_csv(os.path.join(DATA_DIR, "drugs.csv")), "gwas_mr": _load_csv(os.path.join(DATA_DIR, "gwas_mr.csv")), "cmap": _load_csv(os.path.join(DATA_DIR, "cmap_signatures.csv")), "faers": _load_csv(os.path.join(DATA_DIR, "faers_rwe.csv")), "target": _load_csv(os.path.join(DATA_DIR, "target_pathway.csv")), } def _index(rows: list[dict[str, str]], key: str = "drug_id") -> dict[str, dict[str, str]]: return {r[key]: r for r in rows} # --------------------------------------------------------------------------- # Statistical primitives (no rpy2; reproduces Wald ratio + IVW + Cochran Q) # --------------------------------------------------------------------------- def wald_ratio(beta_x: float, beta_y: float, se_y: float, se_x: float) -> tuple[float, float]: """Single-SNP MR Wald ratio with first-order delta-method SE. β_MR = β_y / β_x; SE(β_MR) ≈ sqrt( SE_y²/β_x² + β_y² SE_x² / β_x⁴ ) """ if beta_x == 0: return float("nan"), float("inf") bmr = beta_y / beta_x var = (se_y ** 2) / (beta_x ** 2) + (beta_y ** 2 * se_x ** 2) / (beta_x ** 4) return bmr, math.sqrt(max(var, 0.0)) def ivw_meta(beta_mrs: list[float], ses: list[float]) -> tuple[float, float, float]: """Inverse-variance-weighted meta-analytic MR estimate + Cochran's Q. Returns (β_IVW, SE_IVW, Q).""" if not beta_mrs: return float("nan"), float("inf"), float("nan") weights = [1.0 / (s * s) if s > 0 else 0.0 for s in ses] sw = sum(weights) if sw == 0: return float("nan"), float("inf"), float("nan") beta_ivw = sum(b * w for b, w in zip(beta_mrs, weights)) / sw se_ivw = math.sqrt(1.0 / sw) # Cochran's Q (heterogeneity) Q = sum(w * (b - beta_ivw) ** 2 for b, w in zip(beta_mrs, weights)) return beta_ivw, se_ivw, Q def spearman_rho(a: list[float], b: list[float]) -> float: """Spearman rank correlation (ties handled with average rank).""" if len(a) != len(b) or len(a) < 2: return 0.0 def _rank(xs: list[float]) -> list[float]: order = sorted(range(len(xs)), key=lambda i: xs[i]) ranks = [0.0] * len(xs) i = 0 while i < len(xs): j = i while j + 1 < len(xs) and xs[order[j + 1]] == xs[order[i]]: j += 1 avg = (i + j) / 2.0 + 1.0 for k in range(i, j + 1): ranks[order[k]] = avg i = j + 1 return ranks ra, rb = _rank(a), _rank(b) n = len(a) mean_a = sum(ra) / n mean_b = sum(rb) / n num = sum((ra[i] - mean_a) * (rb[i] - mean_b) for i in range(n)) den_a = math.sqrt(sum((ra[i] - mean_a) ** 2 for i in range(n))) den_b = math.sqrt(sum((rb[i] - mean_b) ** 2 for i in range(n))) if den_a == 0 or den_b == 0: return 0.0 return num / (den_a * den_b) # --------------------------------------------------------------------------- # 5-axis scoring (each axis normalized to 0~100) # --------------------------------------------------------------------------- # T2DM "reference" disease signature direction per tissue (higher = more T2DM-like) T2DM_DISEASE_SIG = { "liver": 1.0, "muscle": 1.0, "adipose": 1.0, "pancreas": 1.0, } def score_mr(row: dict[str, str]) -> tuple[float, dict[str, Any]]: """Score MR axis. Lower β_outcome_t2dm + higher F-stat + more SNPs = stronger protective effect → higher score.""" beta_x = float(row["beta_exposure"]) se_x = float(row["se_exposure"]) beta_y = float(row["beta_outcome_t2dm"]) se_y = float(row["se_outcome"]) beta_y_hba1c = float(row["beta_outcome_hba1c"]) se_y_hba1c = float(row["se_outcome_hba1c"]) f_stat = float(row["f_statistic"]) n_snps = int(row["n_snps"]) # Wald ratios (T2DM + HbA1c) — naive meta to demonstrate IVW bmr_t2dm, se_t2dm = wald_ratio(beta_x, beta_y, se_y, se_x) bmr_hba1c, se_hba1c = wald_ratio(beta_x, beta_y_hba1c, se_y_hba1c, se_x) beta_ivw, se_ivw, Q = ivw_meta([bmr_t2dm, bmr_hba1c], [se_t2dm, se_hba1c]) # Score: more negative β_ivw → better; weight by sqrt(F) and log(n_snps) raw = -beta_ivw * math.sqrt(max(f_stat, 1.0)) * math.log1p(n_snps) # Normalize: empirically map raw [−6, +12] → [0, 100] score = max(0.0, min(100.0, (raw + 6.0) * 100.0 / 18.0)) return score, { "beta_ivw_t2dm_hba1c": round(beta_ivw, 4), "se_ivw": round(se_ivw, 4), "cochran_Q": round(Q, 4), "f_statistic": f_stat, "n_snps": n_snps, "wald_t2dm": round(bmr_t2dm, 4), "wald_hba1c": round(bmr_hba1c, 4), } def score_cmap(row: dict[str, str]) -> tuple[float, dict[str, Any]]: """Score CMAP axis: drug signature ↔ disease signature inverse correlation.""" drug_sig = [ float(row["signature_corr_t2dm_liver"]), float(row["signature_corr_t2dm_muscle"]), float(row["signature_corr_t2dm_adipose"]), float(row["signature_corr_t2dm_pancreas"]), ] disease_sig = [T2DM_DISEASE_SIG["liver"], T2DM_DISEASE_SIG["muscle"], T2DM_DISEASE_SIG["adipose"], T2DM_DISEASE_SIG["pancreas"]] rho = spearman_rho(drug_sig, disease_sig) mean_corr = sum(drug_sig) / len(drug_sig) # Reversal: more negative mean_corr → better; map [-0.7, +0.3] → [100, 0] score = max(0.0, min(100.0, (0.3 - mean_corr) * 100.0)) return score, { "spearman_rho_drug_vs_disease": round(rho, 4), "mean_tissue_correlation": round(mean_corr, 4), "per_tissue": { "liver": drug_sig[0], "muscle": drug_sig[1], "adipose": drug_sig[2], "pancreas": drug_sig[3], }, } def score_target(row: dict[str, str]) -> tuple[float, dict[str, Any]]: """Score target-pathway concordance axis using -log10(p) of GWAS+KEGG overlap and a +/- adjustment for IMPC KO phenotype.""" p_gwas = float(row["t2dm_gwas_overlap_p"]) p_kegg = float(row["kegg_pathway_overlap_p"]) p_combined_log = -math.log10(p_gwas) + -math.log10(p_kegg) glucose_pheno = row["impc_ko_glucose_phenotype"] insulin_pheno = row["impc_ko_insulin_phenotype"] pheno_bonus = 0.0 if "decreased_glucose" in glucose_pheno or "improved" in insulin_pheno: pheno_bonus += 10 if "strongly_improved" in insulin_pheno or "decreased_glucose" == glucose_pheno: pheno_bonus += 5 if "increased_glucose" in glucose_pheno or "decreased_sensitivity" in insulin_pheno: pheno_bonus -= 15 if "strongly_increased" in glucose_pheno: pheno_bonus -= 10 # Map [0, 16] → [0, 80], add bonus score = max(0.0, min(100.0, p_combined_log * 5.0 + pheno_bonus + 20.0)) return score, { "t2dm_gwas_overlap_p": p_gwas, "kegg_pathway_overlap_p": p_kegg, "impc_glucose_phenotype": glucose_pheno, "impc_insulin_phenotype": insulin_pheno, "fisher_combined_neg_log10p": round(p_combined_log, 4), } def score_rwe(row: dict[str, str]) -> tuple[float, dict[str, Any]]: """Score RWE disproportionality axis. RR/ROR/PRR/EBGM < 1 → protective.""" rr = float(row["rr"]) ror = float(row["ror"]) prr = float(row["prr"]) ebgm = float(row["ebgm"]) mean_signal = (rr + ror + prr + ebgm) / 4.0 # Map [0.5, 2.5] → [100, 0] score = max(0.0, min(100.0, (2.5 - mean_signal) * 50.0)) return score, { "rr": rr, "ror": ror, "prr": prr, "ebgm": ebgm, "kcd_e11_signal_kor": row["kcd_e11_signal_kor"], "exposure_n": int(row["exposure_n"]), "t2dm_event_n": int(row["t2dm_event_n"]), } def score_viability(drug_row: dict[str, str]) -> tuple[float, dict[str, Any]]: """Score viability axis (generic·patent·KFDA·HIRA).""" s = 0 s += 25 if drug_row["generic_status"] == "generic" else 10 s += 25 if drug_row["patent_status"] == "expired" else 5 s += 25 if drug_row["kfda_approval"] == "approved" else 0 if drug_row["hira_coverage"] == "covered": s += 25 elif drug_row["hira_coverage"] == "partial": s += 15 else: s += 5 return float(s), { "generic": drug_row["generic_status"], "patent": drug_row["patent_status"], "kfda": drug_row["kfda_approval"], "hira": drug_row["hira_coverage"], } # --------------------------------------------------------------------------- # Composite ranking # --------------------------------------------------------------------------- DEFAULT_WEIGHTS = {"mr": 0.30, "cmap": 0.20, "target": 0.20, "rwe": 0.20, "viability": 0.10} def compute_ranking(data: dict[str, list[dict[str, str]]], weights: dict[str, float] | None = None ) -> list[dict[str, Any]]: weights = weights or DEFAULT_WEIGHTS drugs = _index(data["drugs"]) gwas_mr = _index(data["gwas_mr"]) cmap = _index(data["cmap"]) faers = _index(data["faers"]) target = _index(data["target"]) out = [] for drug_id, drug_row in drugs.items(): # Reference (metformin) is informative but excluded from ranking is_reference = drug_id == "DR001" mr_s, mr_det = score_mr(gwas_mr[drug_id]) if drug_id in gwas_mr else (0.0, {}) cmap_s, cmap_det = score_cmap(cmap[drug_id]) if drug_id in cmap else (0.0, {}) tgt_s, tgt_det = score_target(target[drug_id]) if drug_id in target else (0.0, {}) rwe_s, rwe_det = score_rwe(faers[drug_id]) if drug_id in faers else (0.0, {}) via_s, via_det = score_viability(drug_row) composite = ( weights["mr"] * mr_s + weights["cmap"] * cmap_s + weights["target"] * tgt_s + weights["rwe"] * rwe_s + weights["viability"] * via_s ) out.append({ "drug_id": drug_id, "drug_name_en": drug_row["drug_name_en"], "drug_name_kr": drug_row["drug_name_kr"], "atc_code": drug_row["atc_code"], "drugbank_id": drug_row["drugbank_id"], "original_indication": drug_row["original_indication"], "target_gene": drug_row["target_gene"], "target_pathway": drug_row["target_pathway"], "scores": { "mr": round(mr_s, 2), "cmap": round(cmap_s, 2), "target": round(tgt_s, 2), "rwe": round(rwe_s, 2), "viability": round(via_s, 2), "composite": round(composite, 2), }, "details": { "mr": mr_det, "cmap": cmap_det, "target": tgt_det, "rwe": rwe_det, "viability": via_det, }, "is_reference": is_reference, "drug_record": drug_row, }) # Exclude reference from main ranking but keep at end (for context) ranked = sorted([o for o in out if not o["is_reference"]], key=lambda r: r["scores"]["composite"], reverse=True) for i, r in enumerate(ranked, start=1): r["rank"] = i # Append reference (metformin) marked ref = [o for o in out if o["is_reference"]] for r in ref: r["rank"] = 0 # 0 = reference return ranked + ref # --------------------------------------------------------------------------- # Pretty printers # --------------------------------------------------------------------------- def print_table(rows: list[dict[str, Any]], top: int) -> None: print() print(f"DMRepurpRank-Kor — Top {top} 비당뇨 → 당뇨 재포지셔닝 후보") print("=" * 110) header = ( f"{'순위':>4} {'약물(KR)':<14} {'약물(EN)':<22} {'ATC':<10} " f"{'MR':>6} {'CMAP':>6} {'TGT':>6} {'RWE':>6} {'VIA':>6} {'COMP':>7}" ) print(header) print("-" * 110) shown = 0 for r in rows: if r.get("is_reference"): continue if shown >= top: break s = r["scores"] print( f"{r['rank']:>4} {r['drug_name_kr']:<14} {r['drug_name_en']:<22} " f"{r['atc_code']:<10} " f"{s['mr']:>6.1f} {s['cmap']:>6.1f} {s['target']:>6.1f} " f"{s['rwe']:>6.1f} {s['viability']:>6.1f} {s['composite']:>7.2f}" ) shown += 1 print("-" * 110) # Reference line for r in rows: if r.get("is_reference"): s = r["scores"] print( f"{'REF':>4} {r['drug_name_kr']:<14} {r['drug_name_en']:<22} " f"{r['atc_code']:<10} " f"{s['mr']:>6.1f} {s['cmap']:>6.1f} {s['target']:>6.1f} " f"{s['rwe']:>6.1f} {s['viability']:>6.1f} {s['composite']:>7.2f} ← 당뇨 표준치료 (참조)" ) print() print(f"⚠️ {DISCLAIMER}") print() def format_card(r: dict[str, Any]) -> str: s = r["scores"] d = r["details"] mr = d["mr"] cmap = d["cmap"] tgt = d["target"] rwe = d["rwe"] via = d["viability"] drug = r["drug_record"] pheno_kr = "" glu = tgt.get("impc_glucose_phenotype", "") ins = tgt.get("impc_insulin_phenotype", "") if "decreased_glucose" in glu: pheno_kr += "포도당 ↓ " if "improved" in ins: pheno_kr += "/ 인슐린 감수성 ↑" if "increased_glucose" in glu: pheno_kr += "포도당 ↑ (역방향 시그널) " narrative = ( f"{r['drug_name_kr']}({r['drug_name_en']})은(는) 현재 " f"{drug['original_indication']} 적응증으로 사용 중이며, " f"표적 {r['target_gene']} ({r['target_pathway']}) 활성이 T2DM/HbA1c " f"표현형에 미치는 MR-기반 인과 추정치는 β_IVW={mr.get('beta_ivw_t2dm_hba1c')}, " f"F={mr.get('f_statistic')}, n_SNP={mr.get('n_snps')} 수준으로 관찰됨. " f"CMAP 4-tissue 평균 reversal correlation은 ρ={cmap.get('mean_tissue_correlation')} " f"(disease signature와 음의 상관 → reversal). " f"FAERS/HIRA disproportionality는 RR={rwe.get('rr')} (ROR={rwe.get('ror')}, " f"PRR={rwe.get('prr')}, EBGM={rwe.get('ebgm')}), " f"KCD E11.x 시그널: {rwe.get('kcd_e11_signal_kor')}. " f"IMPC KO 표현형: {pheno_kr or '특기 없음'}." ) grant_block = ( "## KDDF / IIT / Grant proposal template suggestion\n" f"- 제안 grant 후보: KDDF 재포지셔닝 트랙, 국가신약개발사업단(KDDF), " f"보건복지부 의료기술연구개발(R&D), 한국연구재단 중견연구\n" f"- 제안 study design: 1) Korean cohort PoC (HIRA claims + KoGES GWAS proxy) " f"target {r['target_gene']} stratified, 2) 12주 randomized open-label PoC trial " f"(primary: ΔHbA1c, secondary: HOMA-IR / fasting glucose), " f"3) safety: KFDA-approved {drug['kfda_approval']}, HIRA coverage {drug['hira_coverage']}\n" f"- 산학협력: 순천향대 부천병원(닥터앤서 3.0 컨소시엄) + KoGES 코호트\n" f"- 산업적 관점: generic={drug['generic_status']}, patent={drug['patent_status']} → " f"{'low-cost 재포지셔닝 적합' if drug['patent_status']=='expired' else 'IP 회피 설계 필요'}" ) card = ( f"# {r['drug_name_kr']} ({r['drug_name_en']}) 당뇨 재포지셔닝 가설\n" f"- 기존 적응증: {drug['original_indication']}\n" f"- Target: {r['target_gene']} ({r['target_pathway']})\n" f"- ATC: {r['atc_code']} | DrugBank: {r['drugbank_id']}\n\n" f"## 5축 점수 (각 0~100)\n" f"- **MR (Mendelian randomization)**: {s['mr']:.1f}/100 — " f"β_IVW={mr.get('beta_ivw_t2dm_hba1c')}, F={mr.get('f_statistic')}, " f"n_SNP={mr.get('n_snps')}, Q={mr.get('cochran_Q')}\n" f"- **CMAP transcriptomic reversal**: {s['cmap']:.1f}/100 — " f"mean ρ={cmap.get('mean_tissue_correlation')} " f"(liver={cmap.get('per_tissue',{}).get('liver')}, " f"muscle={cmap.get('per_tissue',{}).get('muscle')}, " f"adipose={cmap.get('per_tissue',{}).get('adipose')}, " f"pancreas={cmap.get('per_tissue',{}).get('pancreas')})\n" f"- **Target-pathway concordance**: {s['target']:.1f}/100 — " f"GWAS overlap p={tgt.get('t2dm_gwas_overlap_p')}, " f"KEGG overlap p={tgt.get('kegg_pathway_overlap_p')}, " f"IMPC KO glucose={tgt.get('impc_glucose_phenotype')}, " f"insulin={tgt.get('impc_insulin_phenotype')}\n" f"- **RWE disproportionality**: {s['rwe']:.1f}/100 — " f"RR={rwe.get('rr')}, ROR={rwe.get('ror')}, PRR={rwe.get('prr')}, " f"EBGM={rwe.get('ebgm')}, KCD: {rwe.get('kcd_e11_signal_kor')}, " f"n_exposure={rwe.get('exposure_n')}\n" f"- **Viability**: {s['viability']:.1f}/100 — " f"generic={via.get('generic')}, patent={via.get('patent')}, " f"KFDA={via.get('kfda')}, HIRA={via.get('hira')}\n\n" f"## Composite: {s['composite']:.2f}/100 (Top {r.get('rank','-')})\n\n" f"## 가설 narrative\n{narrative}\n\n" f"{grant_block}\n\n" f"## ⚠️ 디스클레이머\n{DISCLAIMER}\n" ) return card # --------------------------------------------------------------------------- # Weight parser # --------------------------------------------------------------------------- def parse_weights(spec: str) -> dict[str, float]: out = dict(DEFAULT_WEIGHTS) if not spec: return out parts = [p.strip() for p in spec.split(",") if p.strip()] for p in parts: if "=" not in p: raise ValueError(f"잘못된 가중치 형식: {p}") k, v = p.split("=", 1) k = k.strip().lower() if k not in DEFAULT_WEIGHTS: raise ValueError(f"알 수 없는 axis: {k} (허용: {list(DEFAULT_WEIGHTS.keys())})") out[k] = float(v) # Renormalize so they sum to 1 total = sum(out.values()) if total <= 0: raise ValueError("가중치 합이 0보다 커야 합니다.") return {k: v / total for k, v in out.items()} # --------------------------------------------------------------------------- # Summary # --------------------------------------------------------------------------- def print_summary() -> None: drug_n = len(load_all()["drugs"]) body = ( "\n" "DMRepurpRank-Kor (디엠리퍼프랭크코어)\n" "─────────────────────────────────────\n" "도메인 : DM (당뇨병)\n" "카테고리 : 연구 아이디어 생성 / drug repurposing ranker\n" "형식 : Python CLI (argparse)\n" "빌드 일자 : 2026-05-15\n\n" "핵심 기능 (5):\n" " 1) Multi-source ingest + entity 정규화 (DrugBank, OpenTargets, GWAS,\n" " CMAP/L1000, FAERS, HIRA, IMPC, KCD-7/8 등 7+ source 합성 통합)\n" " 2) Mendelian randomization 자동 계산 (Wald ratio, IVW, F-statistic,\n" " Cochran's Q — rpy2 없이 numpy/stdlib로 구현)\n" " 3) Transcriptomic signature reversal (Spearman ρ) + target-pathway\n" " concordance (GWAS+KEGG Fisher combined p)\n" " 4) RWE disproportionality (RR / ROR / PRR / EBGM) + KCD E11.x 시그널\n" " 5) Composite ranking + 한국어 hypothesis card / IIT-grant proposal 출력\n\n" "CLI:\n" " python3 main.py --top 10\n" " python3 main.py --top 10 --weights mr=0.3,cmap=0.2,target=0.2,rwe=0.2,viability=0.1\n" " python3 main.py --card 메트포르민 # 또는 영문 이름 / drug_id\n" " python3 main.py --export-json out.json\n" " python3 main.py --summary\n\n" "데이터 (합성):\n" f" data/drugs.csv {drug_n}개 비당뇨 약물 후보\n" " data/gwas_mr.csv target gene 별 GWAS IV\n" " data/cmap_signatures.csv 4-tissue (liver, muscle, adipose, pancreas) reversal\n" " data/faers_rwe.csv FAERS/HIRA RR/ROR/PRR/EBGM + KCD signal\n" " data/target_pathway.csv GWAS+KEGG overlap + IMPC KO phenotype\n\n" f"⚠️ {DISCLAIMER}\n" ) print(body) # --------------------------------------------------------------------------- # Export # --------------------------------------------------------------------------- def export_json(rows: list[dict[str, Any]], path: str) -> None: # 닥터앤서 3.0 호환 envelope envelope = { "tool": "DMRepurpRank-Kor", "tool_name_kr": "디엠리퍼프랭크코어", "version": "1.0.0", "build_date": "2026-05-15", "domain": "DM", "category": "research_idea_generation", "consortium": "닥터앤서 3.0 / 순천향대 부천병원", "disclaimer": DISCLAIMER, "weights": DEFAULT_WEIGHTS, "sources": [ "DrugBank", "OpenTargets", "UK Biobank GWAS", "KoGES GWAS", "CMAP/L1000", "FAERS", "HIRA claims (KIDS-KD)", "IMPC KO", "KCD-7/8", ], "candidates": [ {k: v for k, v in r.items() if k != "drug_record"} for r in rows ], } with open(path, "w", encoding="utf-8") as f: json.dump(envelope, f, ensure_ascii=False, indent=2) # --------------------------------------------------------------------------- # Lookup helper # --------------------------------------------------------------------------- def find_drug(rows: list[dict[str, Any]], query: str) -> dict[str, Any] | None: q = query.strip().lower() for r in rows: if (r["drug_id"].lower() == q or r["drug_name_en"].lower() == q or r["drug_name_kr"].lower() == q or r["drug_name_en"].lower().startswith(q) or q in r["drug_name_kr"].lower()): return r return None # --------------------------------------------------------------------------- # Main # --------------------------------------------------------------------------- def build_argparser() -> argparse.ArgumentParser: epilog = ( "예시:\n" " python3 main.py --top 10\n" " python3 main.py --top 5 --weights mr=0.4,cmap=0.2,target=0.2,rwe=0.15,viability=0.05\n" " python3 main.py --card 텔미사르탄\n" " python3 main.py --export-json result.json\n" " python3 main.py --summary\n\n" f"⚠️ {DISCLAIMER}" ) p = argparse.ArgumentParser( prog="DMRepurpRank-Kor", description=( "DMRepurpRank-Kor (디엠리퍼프랭크코어) — 비당뇨 약물 → 당뇨 효능 " "재포지셔닝 후보 자동 통합 ranking + 한국어 IIT/grant card 생성 CLI. " f"⚠️ {DISCLAIMER}" ), epilog=epilog, formatter_class=argparse.RawDescriptionHelpFormatter, ) p.add_argument("--top", type=int, default=None, help="상위 N개 후보 ranking 표 출력") p.add_argument("--weights", type=str, default=None, help="가중치 (예: mr=0.3,cmap=0.2,target=0.2,rwe=0.2,viability=0.1)") p.add_argument("--card", type=str, default=None, help="특정 약물 한국어 hypothesis card 출력 (drug_id / EN / KR 이름)") p.add_argument("--export-json", type=str, default=None, metavar="PATH", help="닥터앤서 3.0 호환 JSON으로 export") p.add_argument("--summary", action="store_true", help="도구 개요 요약 출력") return p def main(argv: list[str] | None = None) -> int: parser = build_argparser() args = parser.parse_args(argv) if not (args.top or args.card or args.export_json or args.summary): parser.print_help() return 0 if args.summary: print_summary() return 0 weights = parse_weights(args.weights) if args.weights else DEFAULT_WEIGHTS data = load_all() ranking = compute_ranking(data, weights) did_any = False if args.top: print_table(ranking, args.top) did_any = True if args.card: target = find_drug(ranking, args.card) if target is None: print(f"❌ '{args.card}' 와 일치하는 후보를 찾을 수 없습니다.", file=sys.stderr) return 2 print(format_card(target)) did_any = True if args.export_json: export_json(ranking, args.export_json) print(f"✅ JSON export 완료: {args.export_json} ({len(ranking)}개 후보)") did_any = True return 0 if did_any else 0 if __name__ == "__main__": sys.exit(main())