generate_irefindex_interactome.py

import os
import networkx as nx


WORKDIR = ''
IREFINDEX_FILE = os.path.join(WORKDIR, '9606.mitab.01-22-2018.txt')


def parse_irefindex_alias(alias):
    splits = alias.split('|')
    for split in splits:
        if split.startswith('hgnc:'):
            return split[5:]
        # if split.startswith('icrogid:'):
        #     return split
    return None


def parse_irefindex_file(irefindex_file):
    G = nx.Graph()
    with open(irefindex_file) as rows:
        header = next(rows)
        for row in rows:
            cols = row.strip().split('\t')
            geneA = parse_irefindex_alias(cols[4])
            geneB = parse_irefindex_alias(cols[5])
            if geneA and geneB:
                G.add_edge(geneA, geneB)
    largest_cc = max(nx.connected_components(G), key=len)
    S = G.subgraph(largest_cc).copy()
    return S


irefindex = parse_irefindex_file(IREFINDEX_FILE)
print(len(irefindex.nodes()), len(irefindex.edges())) # 24711 443720