opentargets-database

Open Targets Platform 数据库

概述

Open Targets Platform 整合了遗传学、基因组学、文献和药物数据库中的证据，系统地为 60,000 多个靶点和 20,000 多种疾病/表型计算靶点-疾病关联评分。公开的 GraphQL API（无需身份验证）提供关联评分、来自 20 多个数据源的证据（GWAS、ClinVar、ChEMBL、药物、通路、小鼠模型、表达数据），以及详细的药物-靶点-疾病关联信息。

适用场景

• 根据总体关联评分和证据明细，为某种疾病筛选治疗靶点
• 查找与目标基因相关的所有疾病及其置信度评分
• 获取某个靶点的已获批和在研药物，以及作用机制和临床阶段
• 评估靶点的成药性和可开发性（小分子、抗体、PROTAC 的可能性）
• 提取某个靶点-疾病组合的遗传关联证据（GWAS 命中、变异到基因的映射）
• 从 FAERS 和文献中探索某个药物靶点的安全性/不良事件数据
• 查询生物活性 IC50/Ki 数据，请使用 chembl-database-bioactivity；查询临床试验详情，请使用 clinicaltrials-database-search

先决条件

• Python 包：requests
• 数据要求：基因符号（HGNC）、Ensembl 基因 ID、疾病 EFO ID 或药物名称
• 环境：需要互联网连接；无需身份验证
• 速率限制：未说明硬性限制；执行大型查询时（>100 个靶点）请合理设置请求间隔

pip install requests

快速入门

import requests

OT_URL = "https://api.platform.opentargets.org/api/v4/graphql"

def ot_query(gql, variables=None):
    r = requests.post(OT_URL, json={"query": gql, "variables": variables or {}})
    r.raise_for_status()
    return r.json()["data"]

# Top disease associations for BRCA1
query = """
query TargetDiseases($ensgId: String!) {
  target(ensemblId: $ensgId) {
    id
    approvedSymbol
    associatedDiseases(page: {index: 0, size: 5}) {
      rows {
        disease { id name }
        score
      }
    }
  }
}
"""
data = ot_query(query, {"ensgId": "ENSG00000012048"})
target = data["target"]
print(f"Target: {target['approvedSymbol']}")
for row in target["associatedDiseases"]["rows"]:
    print(f"  {row['disease']['name']}: {row['score']:.3f}")

核心 API

查询 1：按基因符号查找靶点

搜索靶点并获取基本元数据（Ensembl ID、生物类型、描述）。

import requests

OT_URL = "https://api.platform.opentargets.org/api/v4/graphql"

def ot_query(gql, variables=None):
    r = requests.post(OT_URL, json={"query": gql, "variables": variables or {}})
    r.raise_for_status()
    return r.json()["data"]

# Search by gene symbol
query = """
query SearchTarget($sym: String!) {
  search(queryString: $sym, entityNames: ["target"]) {
    hits {
      id
      name
      entity
      object {
        ... on Target {
          approvedSymbol
          approvedName
          biotype
          functionDescriptions
        }
      }
    }
  }
}
"""
data = ot_query(query, {"sym": "BRCA1"})
for hit in data["search"]["hits"][:3]:
    obj = hit.get("object", {})
    print(f"ID: {hit['id']} | {obj.get('approvedSymbol')} | {obj.get('biotype')}")
    descs = obj.get("functionDescriptions", [])
    if descs:
        print(f"  Function: {descs[0][:120]}")

# Direct lookup by Ensembl ID
query2 = """
query Target($ensgId: String!) {
  target(ensemblId: $ensgId) {
    id approvedSymbol approvedName biotype
    tractability { label modality value }
  }
}
"""
data2 = ot_query(query2, {"ensgId": "ENSG00000141510"})  # TP53
t = data2["target"]
print(f"\n{t['approvedSymbol']} ({t['id']}): {t['biotype']}")
print("Tractability:")
for tr in t.get("tractability", [])[:5]:
    print(f"  {tr['modality']} | {tr['label']}: {tr['value']}")

查询 2：靶点-疾病关联

获取某个靶点与所有相关疾病的关联评分。

import requests, pandas as pd

OT_URL = "https://api.platform.opentargets.org/api/v4/graphql"

def ot_query(gql, variables=None):
    r = requests.post(OT_URL, json={"query": gql, "variables": variables or {}})
    r.raise_for_status()
    return r.json()["data"]

query = """
query Associations($ensgId: String!, $size: Int!) {
  target(ensemblId: $ensgId) {
    approvedSymbol
    associatedDiseases(page: {index: 0, size: $size}, orderByScore: "score") {
      count
      rows {
        disease { id name therapeuticAreas { name } }
        score
        datatypeScores { id score }
      }
    }
  }
}
"""
data = ot_query(query, {"ensgId": "ENSG00000012048", "size": 20})
target = data["target"]
assoc = target["associatedDiseases"]
print(f"{target['approvedSymbol']}: {assoc['count']} associated diseases")

rows = []
for r in assoc["rows"]:
    scores = {d["id"]: d["score"] for d in r.get("datatypeScores", [])}
    rows.append({
        "disease": r["disease"]["name"],
        "disease_id": r["disease"]["id"],
        "overall_score": round(r["score"], 4),
        "genetics": round(scores.get("genetic_association", 0), 3),
        "drugs": round(scores.get("known_drug", 0), 3),
        "literature": round(scores.get("literature", 0), 3),
    })

df = pd.DataFrame(rows)
print(df.head(10).to_string(index=False))

查询 3：疾病-靶点关联

给定一种疾病，获取所有相关靶点并按评分排序。

import requests, pandas as pd

OT_URL = "https://api.platform.opentargets.org/api/v4/graphql"

def ot_query(gql, variables=None):
    r = requests.post(OT_URL, json={"query": gql, "variables": variables or {}})
    r.raise_for_status()
    return r.json()["data"]

query = """
query DiseaseTargets($efoId: String!, $size: Int!) {
  disease(efoId: $efoId) {
    id name
    associatedTargets(page: {index: 0, size: $size}, orderByScore: "score") {
      count
      rows {
        target { id approvedSymbol biotype }
        score
        datatypeScores { id score }
      }
    }
  }
}
"""
# EFO_0000305 = breast carcinoma
data = ot_query(query, {"efoId": "EFO_0000305", "size": 10})
disease = data["disease"]
print(f"Disease: {disease['name']}")
print(f"Total associated targets: {disease['associatedTargets']['count']}")

for row in disease["associatedTargets"]["rows"][:5]:
    t = row["target"]
    print(f"  {t['approvedSymbol']:12s} score={row['score']:.3f} biotype={t['biotype']}")

查询 4：靶点的已知药物

获取已获批和在研药物、作用机制及临床阶段。

import requests, pandas as pd

OT_URL = "https://api.platform.opentargets.org/api/v4/graphql"

def ot_query(gql, variables=None):
    r = requests.post(OT_URL, json={"query": gql, "variables": variables or {}})
    r.raise_for_status()
    return r.json()["data"]

query = """
query KnownDrugs($ensgId: String!) {
  target(ensemblId: $ensgId) {
    approvedSymbol
    drugAndClinicalCandidates {
      count
      rows {
        maxClinicalStage
        drug { id name drugType maximumClinicalStage mechanismsOfAction { rows { mechanismOfAction } } }
        diseases { disease { id name } }
      }
    }
  }
}
"""
data = ot_query(query, {"ensgId": "ENSG00000146648"})  # EGFR
target = data["target"]
drugs_data = target["drugAndClinicalCandidates"]
print(f"{target['approvedSymbol']}: {drugs_data['count']} drug-indication pairs")

rows = []
for r in drugs_data["rows"]:
    drug = r["drug"]
    moa = drug.get("mechanismsOfAction") or {}
    moa_first = (moa.get("rows") or [{}])[0].get("mechanismOfAction")
    first_disease = (r.get("diseases") or [{}])[0].get("disease") or {}
    rows.append({
        "drug": drug["name"],
        "type": drug["drugType"],
        "maxClinicalStage": r["maxClinicalStage"],
        "approved": drug["maximumClinicalStage"] == "PHASE_4",
        "indication": first_disease.get("name", "n/a"),
        "mechanism": moa_first,
    })

df = pd.DataFrame(rows).drop_duplicates(subset=["drug", "indication"])
print(df.head(10).to_string(index=False))

查询 5：特定靶点-疾病组合的证据

获取某个靶点-疾病组合的详细证据记录（GWAS、ClinVar、文献）。

import requests

OT_URL = "https://api.platform.opentargets.org/api/v4/graphql"

def ot_query(gql, variables=None):
    r = requests.post(OT_URL, json={"query": gql, "variables": variables or {}})
    r.raise_for_status()
    return r.json()["data"]

query = """
query Evidence($ensgId: String!, $efoId: String!) {
  disease(efoId: $efoId) {
    evidences(
      ensemblIds: [$ensgId]
      enableIndirect: true
      size: 10
      datasourceIds: ["gwas_catalog", "clinvar", "chembl"]
    ) {
      count
      rows {
        datasourceId
        score
        variantRsId
        studyId
        publicationYear
        clinicalSignificances
      }
    }
  }
}
"""
data = ot_query(query, {"ensgId": "ENSG00000012048", "efoId": "EFO_0000305"})
evidences = data["disease"]["evidences"]
print(f"Evidence records: {evidences['count']}")
for ev in evidences["rows"][:5]:
    print(f"  Source: {ev['datasourceId']:20s} | Score: {ev['score']:.3f}")

查询 6：安全性和不良事件

获取某个靶点的已知不良事件和安全性数据。

import requests

OT_URL = "https://api.platform.opentargets.org/api/v4/graphql"

def ot_query(gql, variables=None):
    r = requests.post(OT_URL, json={"query": gql, "variables": variables or {}})
    r.raise_for_status()
    return r.json()["data"]

query = """
query Safety($ensgId: String!) {
  target(ensemblId: $ensgId) {
    approvedSymbol
    safetyLiabilities {
      event
      effects { direction dosing }
      biosamples { tissueLabel cellLabel }
      datasource
    }
  }
}
"""
data = ot_query(query, {"ensgId": "ENSG00000146648"})  # EGFR
target = data["target"]
print(f"Safety liabilities for {target['approvedSymbol']}:")
for s in target.get("safetyLiabilities", [])[:5]:
    print(f"  Event: {s['event']}")
    # 2025 schema: `datasource` is a scalar literature-citation string,
    # not the legacy `datasources[{name, pmid}]` list of objects
    print(f"    Source: {s.get('datasource', 'n/a')}")
    for eff in s.get("effects", []) or []:
        print(f"    Effect: direction={eff.get('direction')} dosing={eff.get('dosing')}")

核心概念

关联评分

Open Targets 使用调和和聚合方法，将多个数据源的证据合并为 0–1 的关联评分。子评分包括：genetic_association、somatic_mutation、known_drug、affected_pathway、literature、RNA_expression、animal_model 等。评分越高，表示证据越多、力度越强。

疾病的 EFO ID

Open Targets 使用实验因子本体（EFO）标识符表示疾病（例如，乳腺癌的 EFO_0000305）。查询关联之前，请使用 search 查询按疾病名称查找 EFO ID。

常见工作流

工作流 1：疾病靶点优先级排序

目标：给定一种疾病，按总体评分对所有相关靶点排序，并导出证据明细。

import requests, pandas as pd, time

OT_URL = "https://api.platform.opentargets.org/api/v4/graphql"

def ot_query(gql, variables=None):
    r = requests.post(OT_URL, json={"query": gql, "variables": variables or {}})
    r.raise_for_status()
    return r.json()["data"]

def disease_search(name):
    q = 'query S($q:String!){search(queryString:$q,entityNames:["disease"]){hits{id name}}}'
    data = ot_query(q, {"q": name})
    return [(h["id"], h["name"]) for h in data["search"]["hits"][:3]]

def get_top_targets(efo_id, n=50):
    q = """
    query($efoId:String!,$size:Int!){
      disease(efoId:$efoId){
        name
        associatedTargets(page:{index:0,size:$size},orderByScore:"score"){
          count
          rows{
            target{id approvedSymbol biotype}
            score
            datatypeScores { id score }
          }
        }
      }
    }"""
    data = ot_query(q, {"efoId": efo_id, "size": n})
    disease = data["disease"]
    rows = []
    for row in disease["associatedTargets"]["rows"]:
        t = row["target"]
        scores = {d["id"]: round(d["score"], 3) for d in row.get("datatypeScores", [])}
        rows.append({
            "target": t["approvedSymbol"],
            "ensembl_id": t["id"],
            "biotype": t["biotype"],
            "overall_score": round(row["score"], 4),
            **scores
        })
    return disease["name"], pd.DataFrame(rows)

# Step 1: Find EFO ID for disease
candidates = disease_search("non-small cell lung carcinoma")
print("Disease candidates:", candidates)

# Step 2: Get top targets
disease_name, df = get_top_targets("EFO_0003060", n=50)
df.to_csv("target_prioritization.csv", index=False)
print(f"\nTop targets for {disease_name}:")
cols = [c for c in ["target", "overall_score", "genetic_association", "known_drug", "literature", "rna_expression", "somatic_mutation"] if c in df.columns]
print(df[cols].head(10).to_string(index=False))

工作流 2：药物-靶点-疾病关联

目标：针对某个靶点，获取所有药物及其相关适应症和临床阶段。

import requests, pandas as pd

OT_URL = "https://api.platform.opentargets.org/api/v4/graphql"

def ot_query(gql, variables=None):
    r = requests.post(OT_URL, json={"query": gql, "variables": variables or {}})
    r.raise_for_status()
    return r.json()["data"]

query = """
query($ensgId:String!){
  target(ensemblId:$ensgId){
    approvedSymbol
    drugAndClinicalCandidates{
      count
      rows{
        maxClinicalStage
        drug{id name drugType maximumClinicalStage mechanismsOfAction { rows { mechanismOfAction } } }
        diseases { disease { id name } }
      }
    }
  }
}"""

targets = {
    "EGFR": "ENSG00000146648",
    "ERBB2": "ENSG00000141736",
}

all_rows = []
for sym, ensg in targets.items():
    data = ot_query(query, {"ensgId": ensg})
    for row in data["target"]["drugAndClinicalCandidates"]["rows"]:
        drug = row["drug"]
        moa = drug.get("mechanismsOfAction") or {}
        moa_first = (moa.get("rows") or [{}])[0].get("mechanismOfAction")
        first_disease = (row.get("diseases") or [{}])[0].get("disease") or {}
        all_rows.append({
            "target": sym,
            "drug": drug["name"],
            "drug_type": drug["drugType"],
            "maxClinicalStage": row["maxClinicalStage"],
            "approved": drug["maximumClinicalStage"] == "PHASE_4",
            "indication": first_disease.get("name", "n/a"),
            "mechanism": moa_first,
        })

df = pd.DataFrame(all_rows)
df.to_csv("drug_target_matrix.csv", index=False)
print(df.head(10).to_string(index=False))

关键参数

参数	模块	默认值	范围/选项	作用
page.size	关联	10	1–10000	每页记录数
page.index	关联	0	0–N	分页页码
orderByScore	关联	"score"	"score"、组件 ID	按评分对关联排序
datasourceIds	证据	所有来源	数据源 ID 列表	按来源筛选证据
enableIndirect	证据	false	true/false	包含子级疾病证据
entityNames	搜索	所有	["target"]、["disease"]	筛选搜索实体类型

最佳实践

1. 疾病使用 EFO ID：疾病名称可能不同；运行关联查询前，始终使用 search 查询获取规范的 EFO ID，以避免名称匹配问题。

2. 分页获取完整结果集：默认页面大小为 10；使用 page.size: 10000 获取完整结果，但请注意，这可能会返回很大的数据负载。

3. 按 datatypeScores 筛选：验证遗传靶点时，筛选 genetic_association 子评分 > 0.1；重新定位药物时，优先考虑 known_drug 子评分。

4. 在证据查询中使用 enableIndirect: true，以纳入疾病亚型（EFO 层级中的子项）的证据。

5. 缓存 GraphQL 响应：Open Targets 数据每季度更新；分析期间缓存响应，以避免重复调用 API。

常见配方

配方：将疾病名称转换为 EFO ID

使用场景：将疾病名称解析为关联查询所需的 EFO ID。

import requests

OT_URL = "https://api.platform.opentargets.org/api/v4/graphql"

query = """
query($q: String!) {
  search(queryString: $q, entityNames: ["disease"]) {
    hits { id name score }
  }
}"""
r = requests.post(OT_URL, json={"query": query, "variables": {"q": "breast cancer"}})
for hit in r.json()["data"]["search"]["hits"][:5]:
    print(f"{hit['id']}: {hit['name']} (score={hit['score']:.3f})")

配方：评估靶点可成药性

使用场景：评估靶点是否适用于小分子、抗体或 PROTAC。

import requests

OT_URL = "https://api.platform.opentargets.org/api/v4/graphql"

query = """
query($ensgId: String!) {
  target(ensemblId: $ensgId) {
    approvedSymbol
    tractability { label modality value }
  }
}"""
r = requests.post(OT_URL, json={"query": query, "variables": {"ensgId": "ENSG00000141510"}})
t = r.json()["data"]["target"]
print(f"Tractability for {t['approvedSymbol']}:")
for tr in t.get("tractability", []):
    if tr["value"]:
        print(f"  [{tr['modality']}] {tr['label']}")

配方：疾病的已批准药物

使用场景：查找具有 4 期证据的疾病所有已批准药物。

import requests, pandas as pd

OT_URL = "https://api.platform.opentargets.org/api/v4/graphql"

query = """
query($efoId: String!) {
  disease(efoId: $efoId) {
    name
    drugAndClinicalCandidates { count rows {
      maxClinicalStage
      drug { name maximumClinicalStage drugType }
    }}
  }
}"""
r = requests.post(OT_URL, json={"query": query, "variables": {"efoId": "EFO_0000305"}})
data = r.json()["data"]["disease"]
approved = [row for row in data["drugAndClinicalCandidates"]["rows"] if row["drug"]["maximumClinicalStage"] == "PHASE_4"]
print(f"Approved drugs for {data['name']}: {len(approved)}")
for row in approved[:5]:
    print(f"  {row['drug']['name']} ({row['drug']['drugType']})  maxStage={row['maxClinicalStage']}")

故障排除

问题	原因	解决方案
HTTP 400，并出现 GraphQL 错误	查询格式错误或字段名称无效	在 https://api.platform.opentargets.org/api/v4/graphql 检查查询是否符合 GraphQL 架构
关联结果中的 rows 为空	EFO ID 未被识别	使用 search 查询查找正确的 EFO ID
未找到靶点	基因符号与 Ensembl ID 不匹配	先使用 search 查询解析 Ensembl ID
大型结果集查询缓慢	page.size 过大	最多设置为 500 行；通过多次请求分页查询
缺少可成药性数据	尚未评估该靶点	并非所有靶点都有可成药性数据；检查 tractability 字段是否非空
drugAndClinicalCandidates 为空	ChEMBL 中没有药物-靶点证据	使用 chembl-database-bioactivity 获取临床前化合物活性

相关技能

• chembl-database-bioactivity — 化合物对靶点的生物活性 IC50/Ki 数据
• clinicaltrials-database-search — 通过 Open Targets 找到的药物的详细临床试验信息
• ensembl-database — 作为 Open Targets 查询输入所需的 Ensembl ID 和变异注释
• string-database-ppi — 用于分析靶点生物学背景的蛋白质-蛋白质相互作用网络

参考资料

• Open Targets Platform — 交互式靶点-疾病浏览器
• Open Targets GraphQL API 文档 — 查询参考和架构
• GraphQL playground — 交互式查询构建器
• Open Targets 数据源 — 证据类型和评分方法