compara-deep-learning/access_data_rest.py

154 lines
4.4 KiB
Python
Raw Permalink Normal View History

2019-07-24 03:27:54 -07:00
import json
import requests
import progressbar
import sys
2019-08-23 02:55:59 -07:00
def update_protein(gene_seq,gene):
t=0
while(t!=2):
2019-07-24 03:27:54 -07:00
try:
server = "https://rest.ensembl.org"
2019-08-23 02:55:59 -07:00
ext = "/sequence/id/"+str(gene)+"?type=protein;multiple_sequences=1"
2019-07-24 03:27:54 -07:00
2019-08-23 02:55:59 -07:00
r = requests.get(server+ext, headers={ "Content-Type" : "application/json"})
2019-07-24 03:27:54 -07:00
if not r.ok:
r.raise_for_status()
sys.exit()
2019-08-23 02:55:59 -07:00
r=r.json()
if len(r)==1:
r=dict(r[0])
gene_seq[gene]=str(r["seq"])
2019-07-24 03:27:54 -07:00
return
else:
2019-08-23 02:55:59 -07:00
maxi=0
maxlen=0
2019-07-24 03:27:54 -07:00
for i in range(len(r)):
2019-08-23 02:55:59 -07:00
m=r[i]
m=dict(m)
if len(m["seq"])>maxlen:
maxi=i
r=dict(r[maxi])
gene_seq[gene]=str(r["seq"])
2019-07-24 03:27:54 -07:00
return
2019-08-23 02:55:59 -07:00
except :
t+=1
#print("\nError:",e)
2019-07-24 03:27:54 -07:00
continue
2019-08-23 02:55:59 -07:00
gene_seq[gene]=""
2019-07-24 03:27:54 -07:00
2019-08-23 02:55:59 -07:00
def update_rest_protein(data,fname):
gids={}
with open("processed/not_found_"+fname+".json","r") as file:
gids=dict(json.load(file))
2019-07-24 03:27:54 -07:00
2019-08-23 02:55:59 -07:00
gids=list(gids.keys())
2019-07-24 03:27:54 -07:00
2019-08-23 02:55:59 -07:00
geneseq={}
2019-07-24 03:27:54 -07:00
server = "https://rest.ensembl.org"
ext = "/sequence/id?type=protein"
2019-08-23 02:55:59 -07:00
headers={ "Content-Type" : "application/json", "Accept" : "application/json"}
2019-07-24 03:27:54 -07:00
2019-08-23 02:55:59 -07:00
for i in progressbar.progressbar(range(0,len(gids)-50,50)):
ids=dict(ids=list(gids[i:i+50]))
2019-07-24 03:27:54 -07:00
while(1):
try:
2019-08-23 02:55:59 -07:00
r = requests.post(server+ext, headers=headers, data=str(json.dumps(ids)))
2019-07-24 03:27:54 -07:00
if not r.ok:
r.raise_for_status()
2019-08-23 02:55:59 -07:00
gs=r.json()
tgs={}
2019-07-24 03:27:54 -07:00
for g in gs:
2019-08-23 02:55:59 -07:00
tgs[g["query"]]=g["seq"]
2019-07-24 03:27:54 -07:00
geneseq.update(tgs)
break
except Exception as e:
2019-08-23 02:55:59 -07:00
print("Error:",e)
2019-07-24 03:27:54 -07:00
continue
data.update(geneseq)
for genes in gids:
try:
2019-08-23 02:55:59 -07:00
_=data[genes]
except:
2019-07-24 03:27:54 -07:00
print(genes)
2019-08-23 02:55:59 -07:00
update_protein(data,genes)
2019-07-24 03:27:54 -07:00
2019-08-23 02:55:59 -07:00
print("Protein Sequences Updated Successfully")
2019-07-24 03:27:54 -07:00
return data
2019-08-23 02:55:59 -07:00
def update(gene_seq,gene):
t=0
while(t!=2):
2019-07-24 03:27:54 -07:00
try:
server = "https://rest.ensembl.org"
2019-08-23 02:55:59 -07:00
ext = "/sequence/id/"+str(gene)+"?type=cds;multiple_sequences=1"
2019-07-24 03:27:54 -07:00
2019-08-23 02:55:59 -07:00
r = requests.get(server+ext, headers={ "Content-Type" : "application/json"})
2019-07-24 03:27:54 -07:00
if not r.ok:
r.raise_for_status()
sys.exit()
2019-08-23 02:55:59 -07:00
r=r.json()
if len(r)==1:
r=dict(r[0])
gene_seq[gene]=str(r["seq"])
2019-07-24 03:27:54 -07:00
return
else:
2019-08-23 02:55:59 -07:00
maxi=0
maxlen=0
2019-07-24 03:27:54 -07:00
for i in range(len(r)):
2019-08-23 02:55:59 -07:00
m=r[i]
m=dict(m)
if len(m["seq"])>maxlen:
maxi=i
r=dict(r[maxi])
gene_seq[gene]=str(r["seq"])
2019-07-24 03:27:54 -07:00
return
2019-08-23 02:55:59 -07:00
except Exception as e:
t+=1
#print("\nError:",e)
2019-07-24 03:27:54 -07:00
continue
2019-08-23 02:55:59 -07:00
gene_seq[gene]=""
2019-07-24 03:27:54 -07:00
2019-08-23 02:55:59 -07:00
def update_rest(data,fname):
gids={}
with open("processed/not_found_"+fname+".json","r") as file:
gids=dict(json.load(file))
2019-07-24 03:27:54 -07:00
2019-08-23 02:55:59 -07:00
gids=list(gids.keys())
2019-07-24 03:27:54 -07:00
2019-08-23 02:55:59 -07:00
geneseq={}
2019-07-24 03:27:54 -07:00
server = "https://rest.ensembl.org"
ext = "/sequence/id?type=cds"
2019-08-23 02:55:59 -07:00
headers={ "Content-Type" : "application/json", "Accept" : "application/json"}
2019-07-24 03:27:54 -07:00
2019-08-23 02:55:59 -07:00
for i in progressbar.progressbar(range(0,len(gids)-50,50)):
ids=dict(ids=list(gids[i:i+50]))
2019-07-24 03:27:54 -07:00
while(1):
try:
2019-08-23 02:55:59 -07:00
r = requests.post(server+ext, headers=headers, data=str(json.dumps(ids)))
2019-07-24 03:27:54 -07:00
if not r.ok:
r.raise_for_status()
2019-08-23 02:55:59 -07:00
gs=r.json()
tgs={}
2019-07-24 03:27:54 -07:00
for g in gs:
2019-08-23 02:55:59 -07:00
tgs[g["query"]]=g["seq"]
2019-07-24 03:27:54 -07:00
geneseq.update(tgs)
break
except Exception as e:
2019-08-23 02:55:59 -07:00
print("Error:",e)
2019-07-24 03:27:54 -07:00
continue
data.update(geneseq)
for genes in gids:
try:
2019-08-23 02:55:59 -07:00
_=data[genes]
except:
2019-07-24 03:27:54 -07:00
print(genes)
2019-08-23 02:55:59 -07:00
update(data,genes)
2019-07-24 03:27:54 -07:00
print("Gene Sequences Updated Successfully")
2019-08-23 02:55:59 -07:00
return data