jaspar_tanimoto.py

#! /usr/bin/env python
import glob
from collections import Counter
prefix = 'gene_'

def get_files_names(glob_pattern = '*.txt'):
    return(glob.glob(glob_pattern))

def get_jaspar_bits(files_names):
    files_refs = map(lambda x: open(x), files_names)
    jaspar_bits = map(lambda x: map(lambda y: y.split()[0], x.readlines()), files_refs)
    map(lambda x: x.close(), files_refs)
    return(jaspar_bits)

def get_tfbs_list(tfbs_list_file = 'tfbs_list'):
    with  open(tfbs_list_file) as fr:
        return (map(lambda x: x.strip(), fr.readlines()))

def jaspar_tfbs_count(jaspar_bits, tfbs_list, files_names):
    counts = map(
        lambda x: map(lambda y: x.setdefault(y, 0), tfbs_list),
        map(lambda x: Counter(x), jaspar_bits))
    return(map(
        lambda x: [x[0]] + x[1],
        zip(map(lambda x: prefix + x.replace('.txt', ''), files_names), counts)))

def format_count_output(tfbs_count):
    lines = map(lambda x: reduce(lambda y, z:  '{0}, {1}'.format(y,z), x), tfbs_count)
    return(reduce(lambda x, y: '{0}\n{1}'.format(x, y), lines))

def count_to_fingerprint(count):
    bits = map(lambda z: '1' * min(int(z), 10) + '0' * max(10 - int(z), 0), count)
    return(reduce(lambda y, z: y + z, bits))

def get_fingerprint(tfbs_count):
    return(map(lambda x: [x[0], count_to_fingerprint(x[1:])],  tfbs_count))

def compute_tanimoto(first, second):
    zipped = zip(map(lambda x: int(x), first), map(lambda y: int(y), second))
    return(round(
        float(sum(map(lambda x: x[0] & x[1], zipped)))  / 
        float(sum(map(lambda x: x[0] | x[1], zipped))), 2))

def get_tanimoto(fingerprint, fingerprint_list):
    return(map(lambda x: compute_tanimoto(fingerprint, x), fingerprint_list))

def fingerprints_to_tanimoto(genes_with_fingerprints):
    fingerprints = map(lambda x: x[1], genes_with_fingerprints)
    return(map(lambda x: [x[0]] + get_tanimoto(x[1], fingerprints), genes_with_fingerprints))

def format_tanimoto_output(genes_with_tanimoto):
    genes = reduce(lambda x, y: '{0},{1}'.format(x, y),  map(lambda x: x[0], genes_with_tanimoto))
    lines = map(lambda x: reduce(lambda y, z: '{0},{1}'.format(y, z), x), genes_with_tanimoto)
    return(reduce(lambda x, y: '{0}\n{1}'.format(x, y), [genes] + lines))


if __name__ == '__main__':
    files_names = get_files_names()
    jaspar_bits = get_jaspar_bits(files_names)
    tfbs_list = get_tfbs_list()
    tfbs_count = jaspar_tfbs_count(
            jaspar_bits=jaspar_bits,
            tfbs_list=tfbs_list,
            files_names=files_names)
    genes_fingerprints = get_fingerprint(tfbs_count)
    genes_with_tanimoto = fingerprints_to_tanimoto(genes_fingerprints)
    print(format_tanimoto_output(genes_with_tanimoto))