In [1]:

    

import os
import gzip
import re
import json

import pandas


    

In [2]:

    

# # Download human entrez gene information
# url = 'ftp://ftp.ncbi.nih.gov/gene/DATA/GENE_INFO/Mammalia/Homo_sapiens.gene_info.gz'
# ! wget --timestamping --directory-prefix download/ $url


    

In [3]:

    

# Read Entrez info dataset
path = os.path.join('download', 'Homo_sapiens.gene_info.gz')

with gzip.open(path, 'rt') as read_file:
    matches = re.match(r'#Format: (.+) \(', next(read_file))
    columns = matches.group(1).split(' ')
    gene_df = pandas.read_table(read_file, names = columns, na_values=['-'])

# Restrict to homo sapiens
gene_df = gene_df.query('tax_id == 9606')

len(gene_df)


    

    
Out[3]:





56352

In [4]:

    

# extract symbols and xrefs
xref_rows = list()
symbol_rows = list()

for i, series in gene_df.iterrows():
    gene_id = series.GeneID
    
    # symbols
    symbol = series.Symbol
    if pandas.notnull(symbol):
        symbol_rows.append((gene_id, 'symbol', symbol))
    
    # synonyms
    synonyms = series.Synonyms
    if pandas.notnull(synonyms):
        for synonym in synonyms.split('|'):
            symbol_rows.append((gene_id, 'synonym', synonym))

    # xrefs
    dbXrefs = series.dbXrefs
    if pandas.notnull(dbXrefs):
        for xref in dbXrefs.split('|'):
            db, ref = xref.split(':', 1)
            xref_rows.append((gene_id, db, ref))

xref_df = pandas.DataFrame(xref_rows, columns=['GeneID', 'resource', 'identifier'])
xref_df.to_csv('data/xrefs-human.tsv', sep='\t', index=False)

symbol_df = pandas.DataFrame(symbol_rows, columns=['GeneID', 'type', 'symbol'])
symbol_df.to_csv('data/symbols-human.tsv', sep='\t', index=False)


    

In [5]:

    

xref_df.head()


    

    
Out[5]:






  

    

      

      
GeneID
      
resource
      
identifier
    
  
  

    

      
0
      
1
      
MIM
      
138670
    
    

      
1
      
1
      
HGNC
      
HGNC:5
    
    

      
2
      
1
      
Ensembl
      
ENSG00000121410
    
    

      
3
      
1
      
HPRD
      
00726
    
    

      
4
      
1
      
Vega
      
OTTHUMG00000183507
    
  

In [6]:

    

symbol_df.head()


    

    
Out[6]:






  

    

      

      
GeneID
      
type
      
symbol
    
  
  

    

      
0
      
1
      
symbol
      
A1BG
    
    

      
1
      
1
      
synonym
      
A1B
    
    

      
2
      
1
      
synonym
      
ABG
    
    

      
3
      
1
      
synonym
      
GAB
    
    

      
4
      
1
      
synonym
      
HYST2477
    
  

In [7]:

    

# save a select columnset
columns = ['tax_id', 'GeneID', 'Symbol', 'chromosome', 'map_location', 'type_of_gene', 'description']
select_df = gene_df[columns]
select_df.to_csv('data/genes-human.tsv', sep='\t', index=False)
select_df.head()


    

    
Out[7]:






  

    

      

      
tax_id
      
GeneID
      
Symbol
      
chromosome
      
map_location
      
type_of_gene
      
description
    
  
  

    

      
0
      
9606
      
1
      
A1BG
      
19
      
19q13.4
      
protein-coding
      
alpha-1-B glycoprotein
    
    

      
1
      
9606
      
2
      
A2M
      
12
      
12p13.31
      
protein-coding
      
alpha-2-macroglobulin
    
    

      
2
      
9606
      
3
      
A2MP1
      
12
      
12p13.31
      
pseudo
      
alpha-2-macroglobulin pseudogene 1
    
    

      
3
      
9606
      
9
      
NAT1
      
8
      
8p22
      
protein-coding
      
N-acetyltransferase 1 (arylamine N-acetyltrans...
    
    

      
4
      
9606
      
10
      
NAT2
      
8
      
8p22
      
protein-coding
      
N-acetyltransferase 2 (arylamine N-acetyltrans...
    
  

In [8]:

    

# Compute dictionaries of symbols to GeneIDs and save as jsons
symbol_df = pandas.read_table('data/symbols-human.tsv')

symbol_to_id = dict()
synonym_to_ids = dict()
for i, row in symbol_df.iterrows():
    symbol = row.symbol
    if pandas.isnull(symbol):
        continue
    gene_id = row.GeneID
    if row.type == 'symbol':
        symbol_to_id[symbol] = gene_id
    if row.type == 'synonym':
        synonym_to_ids.setdefault(symbol, list()).append(gene_id)

with open('data/symbols-human.json', 'w') as write_file:
    json.dump(symbol_to_id, write_file, indent=2, sort_keys=True)
with open('data/synonyms-human.json', 'w') as write_file:
    json.dump(synonym_to_ids, write_file, indent=2, sort_keys=True)


    

In [9]:

    

# Synonyms that uniquely map to GeneIDs
symbol_map = {k: v[0] for k, v in synonym_to_ids.items() if len(v) == 1}

# Override synonyms with symbols
symbol_map.update(symbol_to_id)

with open('data/symbol-map.json', 'w') as write_file:
    json.dump(symbol_map, write_file, indent=2, sort_keys=True)