Spaces:
Sleeping
Sleeping
Update genesis/api_clients/ncbi_api.py
Browse files- genesis/api_clients/ncbi_api.py +53 -22
genesis/api_clients/ncbi_api.py
CHANGED
@@ -1,33 +1,64 @@
|
|
1 |
# genesis/api_clients/ncbi_api.py
|
2 |
import requests
|
3 |
-
import
|
4 |
|
5 |
-
|
6 |
-
BASE_URL = "https://eutils.ncbi.nlm.nih.gov/entrez/eutils"
|
7 |
|
8 |
-
def
|
9 |
-
"""
|
10 |
-
|
|
|
|
|
|
|
11 |
params = {
|
12 |
-
"db":
|
13 |
-
"term":
|
14 |
-
"retmax": max_results,
|
15 |
"retmode": "json",
|
16 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
17 |
}
|
18 |
-
|
19 |
-
|
20 |
-
return
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
21 |
|
22 |
-
|
23 |
-
|
24 |
-
|
|
|
|
|
25 |
params = {
|
26 |
"db": "gene",
|
27 |
-
"id":
|
28 |
-
"retmode": "json"
|
29 |
-
"api_key": NCBI_API_KEY
|
30 |
}
|
31 |
-
|
32 |
-
|
33 |
-
|
|
|
|
1 |
# genesis/api_clients/ncbi_api.py
|
2 |
import requests
|
3 |
+
import xml.etree.ElementTree as ET
|
4 |
|
5 |
+
NCBI_BASE = "https://eutils.ncbi.nlm.nih.gov/entrez/eutils"
|
|
|
6 |
|
7 |
+
def ncbi_search(db: str, term: str, retmax: int = 10):
|
8 |
+
"""
|
9 |
+
Search an NCBI database (pubmed, gene, protein, taxonomy, etc.)
|
10 |
+
and return a list of IDs.
|
11 |
+
"""
|
12 |
+
url = f"{NCBI_BASE}/esearch.fcgi"
|
13 |
params = {
|
14 |
+
"db": db,
|
15 |
+
"term": term,
|
|
|
16 |
"retmode": "json",
|
17 |
+
"retmax": retmax
|
18 |
+
}
|
19 |
+
res = requests.get(url, params=params)
|
20 |
+
res.raise_for_status()
|
21 |
+
data = res.json()
|
22 |
+
return data.get("esearchresult", {}).get("idlist", [])
|
23 |
+
|
24 |
+
def ncbi_fetch(db: str, ids: list, rettype: str = "abstract", retmode: str = "text"):
|
25 |
+
"""
|
26 |
+
Fetch records from an NCBI database by IDs.
|
27 |
+
rettype can be 'abstract', 'fasta', 'gb', etc.
|
28 |
+
"""
|
29 |
+
if not ids:
|
30 |
+
return None
|
31 |
+
|
32 |
+
url = f"{NCBI_BASE}/efetch.fcgi"
|
33 |
+
params = {
|
34 |
+
"db": db,
|
35 |
+
"id": ",".join(ids),
|
36 |
+
"rettype": rettype,
|
37 |
+
"retmode": retmode
|
38 |
}
|
39 |
+
res = requests.get(url, params=params)
|
40 |
+
res.raise_for_status()
|
41 |
+
return res.text
|
42 |
+
|
43 |
+
def get_gene_summary(gene_name: str, organism: str = None):
|
44 |
+
"""
|
45 |
+
Get a gene summary from NCBI Gene.
|
46 |
+
"""
|
47 |
+
query = gene_name
|
48 |
+
if organism:
|
49 |
+
query += f" AND {organism}[Organism]"
|
50 |
|
51 |
+
ids = ncbi_search("gene", query, retmax=1)
|
52 |
+
if not ids:
|
53 |
+
return None
|
54 |
+
|
55 |
+
url = f"{NCBI_BASE}/esummary.fcgi"
|
56 |
params = {
|
57 |
"db": "gene",
|
58 |
+
"id": ids[0],
|
59 |
+
"retmode": "json"
|
|
|
60 |
}
|
61 |
+
res = requests.get(url, params=params)
|
62 |
+
res.raise_for_status()
|
63 |
+
data = res.json()
|
64 |
+
return data.get("result", {}).get(ids[0], {})
|