interpro-database

InterPro 数据库

概述

InterPro 是 EBI 的蛋白质家族、结构域和功能位点综合数据库。它将 13 个成员数据库（Pfam、PANTHER、PIRSF、PRINTS、PROSITE、SMART、CDD、NCBIfam 等）的特征整合为统一的 InterPro 条目，每个条目描述一个同源超家族、结构域、家族、重复序列或保守位点。REST API 地址为 https://www.ebi.ac.uk/interpro/api/，免费且无需身份验证。

适用场景

• 根据 UniProt 登录号识别蛋白质中包含的所有结构域和家族（结构域架构）
• 搜索包含特定结构域或属于特定家族的蛋白质
• 查找编码给定结构域或家族的生物体的分类分布
• 将结构域与 PDB 中的实验 3D 结构关联
• 检查哪些源数据库（Pfam、PANTHER、SMART 等）涵盖某个 InterPro 条目
• 通过关键词（例如“激酶结构域”）发现 InterPro 条目；适用于尚不知道登录号的情况
• 如需获取蛋白质序列、功能注释（GO、通路、活性位点）和 ID 映射，请使用 uniprot-protein-database
• 如需下载结构域比对序列或构建 HMM profile，请直接使用 Pfam；InterPro 是元数据库

前提条件

• Python 软件包：requests、pandas、matplotlib
• 数据要求：UniProt 登录号（例如 P04637）或 InterPro 登录号（例如 IPR011009）
• 环境：需要互联网连接；无需 API 密钥
• 速率限制：未公布硬性限制；批量查询时，请在请求之间使用 time.sleep(1.0)；使用 ?cursor= 或 ?page_size= 分页

pip install requests pandas matplotlib

快速入门

import requests

INTERPRO_BASE = "https://www.ebi.ac.uk/interpro/api"

def interpro_get(path: str, params: dict = None) -> dict:
    """Send a GET request to the InterPro API and return parsed JSON."""
    r = requests.get(
        f"{INTERPRO_BASE}/{path}",
        params=params,
        headers={"Accept": "application/json"},
        timeout=30
    )
    r.raise_for_status()
    return r.json()

# Get domain architecture for TP53 (P04637)
# Note: `protein/uniprot/{acc}/` returns only {metadata}; the entries-per-protein
# data lives at `entry/interpro/protein/uniprot/{acc}/` and is keyed `results`.
data = interpro_get("entry/interpro/protein/uniprot/P04637/")
entries = data.get("results", [])
print(f"InterPro entries for TP53: {data.get('count')}  (this page: {len(entries)})")
for e in entries[:4]:
    m = e["metadata"]
    print(f"  {m['accession']}  {m['type']:<25}  {m['name']}")
# InterPro entries for TP53: 9
#   IPR002117  family                     p53 tumour suppressor family
#   IPR036674  homologous_superfamily     p53-like tetramerisation domain superfamily

核心 API

查询 1：搜索条目

按名称关键词搜索 InterPro 条目，或通过登录号获取特定条目。

import requests

INTERPRO_BASE = "https://www.ebi.ac.uk/interpro/api"

def search_entries(query: str, entry_type: str = None,
                   page_size: int = 20) -> list:
    """Search InterPro entries by keyword; optionally filter by type."""
    params = {"search": query, "page_size": page_size}
    if entry_type:
        params["type"] = entry_type   # family, domain, homologous_superfamily, repeat, site
    r = requests.get(
        f"{INTERPRO_BASE}/entry/interpro/",
        params=params,
        headers={"Accept": "application/json"},
        timeout=30
    )
    r.raise_for_status()
    return r.json().get("results", [])

hits = search_entries("serine kinase", entry_type="domain")
print(f"InterPro domain entries matching 'serine kinase': {len(hits)}")
for h in hits[:5]:
    m = h["metadata"]
    print(f"  {m['accession']}  {m['type']:<10}  {m['name']}")
# InterPro domain entries matching 'serine kinase': 8
#   IPR000719  domain    Protein kinase domain
#   IPR008271  domain    Serine/threonine/tyrosine kinase, active site

# Fetch a specific InterPro entry by accession
r = requests.get(
    f"{INTERPRO_BASE}/entry/interpro/IPR000719/",
    headers={"Accept": "application/json"},
    timeout=30
)
r.raise_for_status()
meta = r.json()["metadata"]
print(f"Accession    : {meta['accession']}")
print(f"Name         : {meta['name']}")
print(f"Type         : {meta['type']}")
print(f"Member DBs   : {list(meta.get('member_databases', {}).keys())}")
go_terms = meta.get("go_terms", [])
print(f"GO terms     : {[g['identifier'] for g in go_terms[:3]]}")
# Accession    : IPR000719
# Name         : Protein kinase domain
# Type         : domain
# Member DBs   : ['pfam', 'smart', 'cdd', 'ncbifam', 'panther']
# GO terms     : ['GO:0004672', 'GO:0005524', 'GO:0006468']

查询 2：蛋白质结构域架构

根据 UniProt 登录号，获取蛋白质中匹配到的所有 InterPro 条目（结构域、家族、位点）。

import requests

INTERPRO_BASE = "https://www.ebi.ac.uk/interpro/api"

def get_protein_domain_architecture(uniprot_acc: str) -> dict:
    """Return all InterPro entry matches for a protein. Uses the
    `entry/interpro/protein/uniprot/{acc}/` endpoint, which returns
    {count, next, previous, results}. Each result has metadata + a
    nested `proteins[0].entry_protein_locations` for the per-protein match."""
    r = requests.get(
        f"{INTERPRO_BASE}/entry/interpro/protein/uniprot/{uniprot_acc}/",
        headers={"Accept": "application/json"},
        timeout=60
    )
    r.raise_for_status()
    return r.json()

data = get_protein_domain_architecture("P04637")   # TP53
results = data.get("results", [])
# Pull length/source from the first match's nested protein record
prot0 = results[0]["proteins"][0] if results and results[0].get("proteins") else {}
print(f"Protein length : {prot0.get('protein_length')}")
print(f"Source DB      : {prot0.get('source_database')}")
print(f"InterPro entries: {data.get('count')}")
for entry in results[:6]:
    m = entry["metadata"]
    # Locations are nested under proteins[0].entry_protein_locations
    locs = entry["proteins"][0].get("entry_protein_locations", []) if entry.get("proteins") else []
    loc_str = ", ".join(
        f"{frag['start']}-{frag['end']}"
        for loc in locs for frag in loc.get("fragments", [])
    )
    print(f"  {m['accession']}  {m['type']:<25}  {m['name'][:35]:<35}  [{loc_str}]")

# Compare domain architectures of two proteins side-by-side
import pandas as pd

def domain_set(uniprot_acc: str) -> set:
    data = get_protein_domain_architecture(uniprot_acc)
    return {e["metadata"]["accession"] for e in data.get("results", [])}

brca1_domains = domain_set("P38398")   # BRCA1
tp53_domains   = domain_set("P04637")  # TP53

shared = brca1_domains & tp53_domains
unique_brca1 = brca1_domains - tp53_domains
unique_tp53  = tp53_domains - brca1_domains
print(f"Shared InterPro entries: {len(shared)}")
print(f"BRCA1-unique           : {len(unique_brca1)}")
print(f"TP53-unique            : {len(unique_tp53)}")

查询 3：条目对应的蛋白质

列出包含特定 InterPro 条目（家族或结构域）的蛋白质。

import requests, time

INTERPRO_BASE = "https://www.ebi.ac.uk/interpro/api"

def get_entry_proteins(interpro_acc: str, reviewed_only: bool = True,
                       page_size: int = 50) -> list:
    """Return proteins (UniProt) containing a given InterPro entry.
    Path order is `/protein/{db}/entry/interpro/{IPR}/` — the inverse
    `entry/interpro/{IPR}/protein/{db}/` times out (408) on large families."""
    db = "reviewed" if reviewed_only else "uniprot"
    r = requests.get(
        f"{INTERPRO_BASE}/protein/{db}/entry/interpro/{interpro_acc}/",
        params={"page_size": page_size},
        headers={"Accept": "application/json"},
        timeout=60
    )
    r.raise_for_status()
    return r.json().get("results", [])

proteins = get_entry_proteins("IPR011009")   # Protein kinase-like domain SF
print(f"Reviewed proteins with IPR011009 (page 1): {len(proteins)}")
for p in proteins[:4]:
    m = p["metadata"]
    # metadata fields: accession, gene, length, name, source_database, source_organism
    print(f"  {m['accession']}  {(m.get('gene') or ''):<8}  "
          f"len={m.get('length', '?')}  "
          f"org={(m.get('source_organism') or {}).get('scientificName', '')[:30]}")

# Paginate all proteins for a family using cursor
def get_all_entry_proteins(interpro_acc: str,
                            reviewed_only: bool = True) -> list:
    INTERPRO_BASE = "https://www.ebi.ac.uk/interpro/api"
    db = "reviewed" if reviewed_only else "uniprot"
    # Path-inverted: /protein/{db}/entry/interpro/{IPR}/ is the working order
    url = f"{INTERPRO_BASE}/protein/{db}/entry/interpro/{interpro_acc}/"
    all_proteins = []
    params = {"page_size": 200}
    while url:
        r = requests.get(url, params=params,
                         headers={"Accept": "application/json"}, timeout=60)
        r.raise_for_status()
        data = r.json()
        all_proteins.extend(data.get("results", []))
        url = data.get("next")
        params = None   # next URL already has params encoded
        if url:
            time.sleep(1.0)
    return all_proteins

proteins = get_all_entry_proteins("IPR000719")   # Protein kinase domain
print(f"Total reviewed proteins with protein kinase domain: {len(proteins)}")

查询 4：条目分类

获取带有指定 InterPro 条目注释的蛋白质的分类分布。

import requests

INTERPRO_BASE = "https://www.ebi.ac.uk/interpro/api"

def get_entry_taxonomy(interpro_acc: str,
                        page_size: int = 50) -> list:
    """Return taxonomic summary for proteins in a given InterPro entry.
    Path-inverted: `/taxonomy/uniprot/entry/interpro/{IPR}/`. Each result
    has `metadata` (taxon: accession=taxId, name, parent, children, rank)
    and `entries[]` (representative protein-match locations for that taxon)."""
    r = requests.get(
        f"{INTERPRO_BASE}/taxonomy/uniprot/entry/interpro/{interpro_acc}/",
        params={"page_size": page_size},
        headers={"Accept": "application/json"},
        timeout=90
    )
    r.raise_for_status()
    return r.json().get("results", [])

# Use a smaller entry (p53 DBD); IPR000719 (kinase) has ~270k taxa and times out.
taxa = get_entry_taxonomy("IPR011615")
print(f"Top taxa for IPR011615 (p53 DNA-binding domain):")
for t in taxa[:8]:
    m = t["metadata"]
    print(f"  taxId={m['accession']:>10}  {m.get('name', ''):<30}  "
          f"rank={m.get('rank') or 'n/a'}")

查询 5：结构整合

获取与 InterPro 条目关联的 PDB 结构。

import requests

INTERPRO_BASE = "https://www.ebi.ac.uk/interpro/api"

def get_entry_structures(interpro_acc: str, page_size: int = 25) -> list:
    """Return PDB structures that include a match to a given InterPro entry.
    Path-inverted: `/structure/pdb/entry/interpro/{IPR}/`. The flat form with
    `?entry_interpro=...` is silently slow / 408s on this resource."""
    r = requests.get(
        f"{INTERPRO_BASE}/structure/pdb/entry/interpro/{interpro_acc}/",
        params={"page_size": page_size},
        headers={"Accept": "application/json"},
        timeout=60
    )
    r.raise_for_status()
    return r.json().get("results", [])

structures = get_entry_structures("IPR011009")   # Protein kinase-like SF
print(f"PDB structures linked to IPR011009 (page 1): {len(structures)}")
for s in structures[:5]:
    m = s["metadata"]
    print(f"  {m['accession'].upper()}  resolution={m.get('resolution', 'N/A')} Å  "
          f"experiment={m.get('experiment_type', 'N/A')}")
# PDB structures linked to IPR011009: ~8,000+
#   1A06  resolution=2.5 Å  experiment=x-ray
#   ...

查询 6：结构域序列获取

下载 InterPro 家族中蛋白质的 FASTA 序列，用于序列比对或系统发育分析。

import requests, time

INTERPRO_BASE = "https://www.ebi.ac.uk/interpro/api"

def get_family_fasta(interpro_acc: str,
                      reviewed_only: bool = True,
                      max_sequences: int = 100) -> str:
    """Retrieve FASTA sequences for proteins in an InterPro entry."""
    db = "reviewed" if reviewed_only else "uniprot"
    proteins = []
    # Path-inverted: protein-list-for-entry is /protein/{db}/entry/interpro/{IPR}/
    url = f"{INTERPRO_BASE}/protein/{db}/entry/interpro/{interpro_acc}/"
    params = {"page_size": min(max_sequences, 200)}
    while url and len(proteins) < max_sequences:
        r = requests.get(url, params=params,
                         headers={"Accept": "application/json"}, timeout=60)
        r.raise_for_status()
        data = r.json()
        proteins.extend(data.get("results", []))
        url = data.get("next") if len(proteins) < max_sequences else None
        params = None
        if url:
            time.sleep(1.0)

    # Fetch FASTA from UniProt for each accession
    accessions = [p["metadata"]["accession"] for p in proteins[:max_sequences]]
    fasta_url = "https://rest.uniprot.org/uniprotkb/stream"
    query = " OR ".join(f"accession:{acc}" for acc in accessions)
    r = requests.get(fasta_url,
                     params={"query": query, "format": "fasta"},
                     timeout=120)
    r.raise_for_status()
    return r.text

fasta = get_family_fasta("IPR000719", reviewed_only=True, max_sequences=20)
seq_count = fasta.count(">")
print(f"FASTA sequences retrieved: {seq_count}")
print(fasta[:300])   # preview first sequence header + start

核心概念

InterPro 条目类型

InterPro 将条目分为五种类型。类型决定了匹配结果所代表的生物学关系：

类型	说明	示例
family	具有共同祖先和功能的同源蛋白质群组	IPR000719（蛋白激酶）
domain	可出现在多种蛋白质背景中的独立结构和功能单元	IPR011009（蛋白激酶样超家族）
homologous_superfamily	结构相似、但序列可能已发生分化的结构域	IPR011993（Pleckstrin 样）
repeat	在蛋白质中多次出现的短重复序列单元	IPR001440（TPR 重复序列）
site	短保守基序：活性位点、结合位点或翻译后修饰位点	IPR008271（丝氨酸/苏氨酸激酶活性位点）

成员数据库层级结构

每个 InterPro 条目都会整合一个或多个成员数据库的特征模型。InterPro 登录号（IPR...）是统一的元条目；成员数据库登录号则指向底层模型：

成员数据库	登录号前缀	建模方法
Pfam	PF	隐马尔可夫模型（profile HMM）
PANTHER	PTHR	系统发育树 + HMM
PIRSF	PIRSF	全长 HMM
PRINTS	PR	指纹基序组
PROSITE	PS	模式和 profile
SMART	SM	与数据库集成的 HMM
CDD	cd	位置特异性评分矩阵（PSSM）
NCBIfam	NF	由 NCBI 整理的 HMM

分页

InterPro API 在集合级别对结果分页。每个响应都包含一个 next URL（结果用尽时为 null）和一个 count 字段。对于大型家族（例如激酶：10,000+ 个蛋白质），请务必使用 next 游标逐页获取。

import requests, time

def iterate_interpro(url: str, page_size: int = 200) -> list:
    """Generic paginator for any InterPro list endpoint."""
    results = []
    params = {"page_size": page_size}
    while url:
        r = requests.get(url, params=params,
                         headers={"Accept": "application/json"}, timeout=60)
        r.raise_for_status()
        data = r.json()
        results.extend(data.get("results", []))
        url = data.get("next")
        params = None
        if url:
            time.sleep(1.0)
    return results

常见工作流

工作流 1：蛋白质集的结构域架构报告

目标：获取一组蛋白质的所有 InterPro 结构域，并生成汇总表，展示每个蛋白质包含哪些结构域。

import requests, time, pandas as pd

INTERPRO_BASE = "https://www.ebi.ac.uk/interpro/api"

def get_domains(uniprot_acc: str) -> list:
    """List InterPro entries for a protein. Uses the
    `entry/interpro/protein/uniprot/{acc}/` endpoint (keyed `results`)."""
    r = requests.get(
        f"{INTERPRO_BASE}/entry/interpro/protein/uniprot/{uniprot_acc}/",
        headers={"Accept": "application/json"}, timeout=60
    )
    if r.status_code == 404:
        return []
    r.raise_for_status()
    data = r.json()
    return [
        {
            "protein": uniprot_acc,
            "accession": e["metadata"]["accession"],
            "name": e["metadata"]["name"],
            "type": e["metadata"]["type"],
            "source_db": list(e["metadata"].get("member_databases", {}).keys()),
        }
        for e in data.get("results", [])
    ]

proteins = ["P04637", "P38398", "Q00987", "P10415"]  # TP53, BRCA1, MDM2, BCL2
rows = []
for acc in proteins:
    rows.extend(get_domains(acc))
    time.sleep(1.0)

df = pd.DataFrame(rows)
print(f"Total domain matches: {len(df)}")
print(df.groupby(["protein", "type"])["accession"].count().unstack(fill_value=0))

# Pivot: proteins × domain accessions
pivot = df[df["type"] == "domain"].pivot_table(
    index="protein", columns="accession", aggfunc="size", fill_value=0
)
pivot.to_csv("domain_architecture_matrix.csv")
print(f"\nDomain × protein matrix: {pivot.shape}")

工作流 2：查找具有 PDB 结构的激酶家族成员

目标：获取激酶结构域家族中具有 PDB 实验结构的蛋白质，并按分辨率排序。

import requests, time, pandas as pd

INTERPRO_BASE = "https://www.ebi.ac.uk/interpro/api"

# Step 1: Get PDB structures linked to the protein kinase-like SF entry.
# Use the path-inverted form; the flat `?entry_interpro=` filter 408s.
r = requests.get(
    f"{INTERPRO_BASE}/structure/pdb/entry/interpro/IPR011009/",
    params={"page_size": 200},
    headers={"Accept": "application/json"}, timeout=60
)
r.raise_for_status()
structures = r.json().get("results", [])
print(f"PDB structures with IPR011009 (kinase-like SF, page 1): {len(structures)}")

rows = []
for s in structures:
    m = s["metadata"]
    rows.append({
        "pdb_id": m["accession"],
        "resolution": m.get("resolution"),
        "experiment": m.get("experiment_type", ""),
        "name": m.get("name", ""),
    })

df = pd.DataFrame(rows)
df = df.dropna(subset=["resolution"]).sort_values("resolution")
print(f"\nTop 10 highest-resolution kinase structures:")
print(df[["pdb_id", "resolution", "experiment", "name"]].head(10).to_string(index=False))
df.to_csv("kinase_structures.csv", index=False)
print(f"\nSaved kinase_structures.csv ({len(df)} X-ray / cryo-EM structures)")

工作流 3：结构域的分类学覆盖率柱状图

目标：可视化各主要界中携带指定 InterPro 结构域的已审校蛋白质数量。

import requests, time
import pandas as pd
import matplotlib.pyplot as plt

INTERPRO_BASE = "https://www.ebi.ac.uk/interpro/api"

def get_taxonomy_counts(interpro_acc: str, page_size: int = 100,
                        max_pages: int = 5) -> pd.DataFrame:
    """Walk the taxonomy results for an InterPro entry. The API does not
    expose a per-taxon protein count at this endpoint — instead each
    paged record is one (taxon × representative protein-match) row.
    Aggregate client-side by taxon name to approximate frequency."""
    rows, url = [], f"{INTERPRO_BASE}/taxonomy/uniprot/entry/interpro/{interpro_acc}/"
    params = {"page_size": page_size}
    for _ in range(max_pages):
        if not url:
            break
        r = requests.get(url, params=params,
                         headers={"Accept": "application/json"}, timeout=90)
        r.raise_for_status()
        data = r.json()
        for t in data.get("results", []):
            m = t["metadata"]
            rows.append({
                "taxon_id": m["accession"],
                "name": m.get("name", ""),
                "rank": m.get("rank") or "",
            })
        url = data.get("next")
        params = None
        if url:
            time.sleep(1.0)
    return pd.DataFrame(rows)

IPR_ACC = "IPR011615"   # p53 DNA-binding domain (smaller; kinase 408s)
df = get_taxonomy_counts(IPR_ACC, max_pages=3)
print(f"Tax entries pulled for {IPR_ACC}: {len(df)}")

# Aggregate by name and take top 15 (each row = one rep. protein-match)
top = (df.groupby("name").size().sort_values(ascending=False).head(15)
       .reset_index(name="rep_matches"))
fig, ax = plt.subplots(figsize=(10, 5))
bars = ax.barh(top["name"], top["rep_matches"], color="#2171B5")
ax.bar_label(bars, fmt="%d", padding=3, fontsize=8)
ax.set_xlabel("Representative protein-matches")
ax.set_title(f"Taxonomic distribution of {IPR_ACC} (p53 DNA-binding domain)")
ax.invert_yaxis()
plt.tight_layout()
plt.savefig(f"{IPR_ACC}_taxonomy.png", dpi=150, bbox_inches="tight")
print(f"Saved {IPR_ACC}_taxonomy.png")

关键参数

参数	端点	默认值	范围 / 选项	作用
search	entry/interpro/	—	任意文本字符串	按条目名称和简称筛选关键词
type	entry/interpro/	所有类型	family、domain、homologous_superfamily、repeat、site	按 InterPro 类型筛选条目
page_size	所有列表端点	20	1–200	每页返回的结果数
entry_interpro	structure/pdb/	—	IPR######	按关联的 InterPro 条目筛选结构
source_database	protein/	—	reviewed、uniprot、trembl	按 UniProt 审校级别筛选蛋白质
reviewed（URL 路径）	entry/{ipr}/{acc}/protein/	uniprot	reviewed、uniprot	仅限 Swiss-Prot 已审校条目，或所有 UniProtKB 条目
relations	entry/interpro/{acc}/	—	contains、contained_by、child_of、parent_of	浏览 InterPro 层级结构
next	所有列表端点	—	响应中的 URL	基于游标分页；使用 next 字段中的完整 URL

最佳实践

1. 整理结构域列表时使用 reviewed 蛋白质：未经审校的 TrEMBL 数据集规模大 5–10 倍，并包含自动预测结果。对于基准测试、家族分析或训练集，请限制为 reviewed（Swiss-Prot）条目，以避免未经审校预测带来的噪声。

2. 将大型分类群或蛋白质列表分块处理：获取蛋白激酶超家族等广泛家族的所有 10,000+ 个蛋白质可能耗时数分钟，并生成很大的响应数据。使用 page_size=200 和 next 游标限制查询；将中间结果保存到磁盘。

3. 分页请求之间添加 time.sleep(1.0)：InterPro API 使用 EBI 共享基础设施，未公布速率限制。批处理脚本中每页暂停 1 秒是安全的最低间隔。

4. 跨数据库查询时优先使用 InterPro 登录号，而非成员数据库登录号：Pfam PF00069 和 PANTHER PTHR24340 都对激酶结构域建模，但覆盖的蛋白质不同。使用上级 InterPro IPR000719，即可通过一次查询获得所有成员数据库匹配结果的并集。

5. 解读 entry_protein_locations 前检查 type：只有 domain、repeat 和 site 条目包含有意义的位置信息。family 和 homologous_superfamily 条目通常覆盖整个蛋白质，其坐标参考价值较低。

常用操作

配方：快速检查蛋白质的结构域

适用场景：给定一个 UniProt 登录号，快速列出它包含的 InterPro 结构域。

import requests

INTERPRO_BASE = "https://www.ebi.ac.uk/interpro/api"

def list_protein_domains(uniprot_acc: str) -> list:
    """Return list of (accession, type, name) tuples for a protein."""
    r = requests.get(
        f"{INTERPRO_BASE}/entry/interpro/protein/uniprot/{uniprot_acc}/",
        headers={"Accept": "application/json"}, timeout=60
    )
    r.raise_for_status()
    return [
        (e["metadata"]["accession"], e["metadata"]["type"], e["metadata"]["name"])
        for e in r.json().get("results", [])
    ]

domains = list_protein_domains("P00533")   # EGFR
print(f"InterPro entries in EGFR (P00533): {len(domains)}")
for acc, etype, name in domains:
    print(f"  {acc}  {etype:<25}  {name}")
# InterPro entries in EGFR (P00533): 10
#   IPR009030  homologous_superfamily   Growth factor receptor, cysteine-rich
#   IPR000719  domain                   Protein kinase domain

配方：查找家族中所有蛋白质及其来源数据库覆盖情况

适用场景：了解每个成员数据库（Pfam、PANTHER、SMART 等）覆盖了某个结构域家族中的多少蛋白质。

import requests, time
import pandas as pd

INTERPRO_BASE = "https://www.ebi.ac.uk/interpro/api"

interpro_acc = "IPR000719"   # Protein kinase domain
r = requests.get(
    f"{INTERPRO_BASE}/entry/interpro/{interpro_acc}/",
    headers={"Accept": "application/json"}, timeout=30
)
r.raise_for_status()
member_dbs = r.json()["metadata"].get("member_databases", {})
print(f"Member databases for {interpro_acc}:")
for db, details in member_dbs.items():
    print(f"  {db}: {details}")

# Visualize member database source breakdown
labels = list(member_dbs.keys())
import matplotlib.pyplot as plt
fig, ax = plt.subplots(figsize=(7, 4))
ax.bar(labels, [1] * len(labels), color="#4472C4")   # presence/absence per DB
ax.set_ylabel("Integrated (1=yes)")
ax.set_title(f"Member databases in {interpro_acc}")
plt.tight_layout()
plt.savefig(f"{interpro_acc}_member_dbs.png", dpi=150, bbox_inches="tight")

配方：获取 InterPro 条目的 GO 术语

适用场景：从结构域关联到 GO 功能注释。

import requests

INTERPRO_BASE = "https://www.ebi.ac.uk/interpro/api"

def get_go_terms_for_entry(interpro_acc: str) -> list:
    """Return GO terms associated with an InterPro entry."""
    r = requests.get(
        f"{INTERPRO_BASE}/entry/interpro/{interpro_acc}/",
        headers={"Accept": "application/json"}, timeout=30
    )
    r.raise_for_status()
    go_terms = r.json()["metadata"].get("go_terms", [])
    return [
        {"id": g["identifier"], "name": g["name"],
         "category": g.get("category", {}).get("name", "")}
        for g in go_terms
    ]

go_terms = get_go_terms_for_entry("IPR000719")
print(f"GO terms for IPR000719 (protein kinase domain): {len(go_terms)}")
for g in go_terms:
    print(f"  {g['id']}  [{g['category'][:2].upper()}]  {g['name']}")
# GO terms for IPR000719 (protein kinase domain): 3
#   GO:0004672  [MO]  protein kinase activity
#   GO:0005524  [MO]  ATP binding
#   GO:0006468  [BI]  protein phosphorylation

故障排除

问题	原因	解决方案
蛋白质查询返回 HTTP 404	InterPro 中找不到该登录号	确认该 UniProt 登录号存在；同工型登录号（P12345-2）可能不会单独编入索引
蛋白质的条目列表为空	该蛋白质没有 InterPro 匹配项（例如，内在无序蛋白）	直接检查 UniProt；并非所有蛋白质都有分类结构域
protein/uniprot/{acc}/ 仅返回 metadata（没有条目）	该端点仅返回 蛋白质 信息；条目匹配项位于其他位置	使用 entry/interpro/protein/uniprot/{acc}/ 并读取 results[] 键
entry/interpro/{IPR}/protein/{db}/ 返回 408 / 挂起	以 entry/... 开头的路径会先执行较慢的联接操作	调换路径顺序：protein/{db}/entry/interpro/{IPR}/
structure/pdb/?entry_interpro={IPR} 超时（408）	同样是联接顺序问题	使用 structure/pdb/entry/interpro/{IPR}/
大型家族的 entry/interpro/{IPR}/taxonomy/uniprot/ 返回 408	同上	使用 taxonomy/uniprot/entry/interpro/{IPR}/；对于非常大的条目（例如 IPR000719 激酶），调换顺序后仍可能返回 408——请改用更具体的子家族条目
条目搜索返回 HTTP 400	查询参数无效或不支持该 type 值	使用以下值之一：family、domain、homologous_superfamily、repeat、site
分页提前停止	在达到预期数量之前，next 已为 null	这是正常情况；所有结果均已返回
大型家族响应非常慢	蛋白质集合包含数千个成员	将 page_size 增至 200；每获取一页后就保存结果
出现 ConnectionError 或 Timeout	网络或服务暂时出现问题	使用指数退避策略重试；EBI 服务偶尔会短暂不可用
缺少成员数据库登录号	条目较新，成员数据库集成尚未完成	查询时使用 InterPro 登录号；成员数据库级别的详细信息会随每次发布更新

相关技能

• uniprot-protein-database — 用于获取蛋白质序列、Swiss-Prot 功能注释（活性位点、PTM、疾病关联）以及进行 ID 映射的 UniProt REST API
• esm-protein-language-model — 为序列生成蛋白质语言模型嵌入；确定蛋白质所属的 InterPro 家族后，此工具很有用
• pdb-database — 按 PDB ID 获取并下载实验测定的 3D 结构；可与通过 InterPro 结构查询发现的结构 ID 交叉参照

参考资料

• InterPro REST API 文档 — 端点参考、筛选器和查询示例
• Blum 等，《Nucleic Acids Research》，2021 — 介绍成员数据库集成的 InterPro 重要论文
• InterPro Web 门户 — 交互式蛋白质结构域浏览器
• Paysan-Lafosse 等，《Nucleic Acids Research》，2023 — 介绍新条目类型和成员数据库的 InterPro 2023 年更新