Source code for deepcpg.motifs

"""Motif analysis."""

from __future__ import division
from __future__ import print_function

from collections import OrderedDict
import re

import pandas as pd


def read_tomtom(path):
    """Read Tomtom output file."""
    d = pd.read_table(path)
    d.rename(columns={'#Query ID': 'Query ID'}, inplace=True)
    d.columns = [x.lower() for x in d.columns]
    d['idx'] = [int(x) for x in d['query id'].str.replace('filter', '')]
    return d


def read_meme_db(meme_db_file):
    """Read MEME database as Pandas DataFrame.

    Parameters
    ----------
    meme_db_file: str
        File name of MEME database.

    Returns
    -------
    :class:`pandas.DataFrame`
        :class:`pandas.DataFrame` with columns 'id', 'protein', 'url'.
    """
    motifs = []
    motif = None
    for line in open(meme_db_file):
        if line.startswith('MOTIF'):
            if motif:
                motifs.append(motif)
                motif = None
            tmp = line.split()[1:]
            if len(tmp) < 2:
                continue
            motif = OrderedDict()
            motif['id'] = tmp[0]
            protein = re.sub(r'\(([^)]+)\)', r'\1', tmp[1])
            motif['protein'] = protein.split('_')[0]
            motif['url'] = ''
        elif motif and line.startswith('URL'):
            motif['url'] = line.split()[1]
    if motif:
        motifs.append(motif)
    for i, motif in enumerate(motifs):
        motifs[i] = pd.DataFrame(motif, index=[0])
    motifs = pd.concat(motifs)
    return motifs


def get_report(filter_stats_file, tomtom_file, meme_motifs):
    """Read and join `filter_stats_file` and `tomtom_file`.

    Used by `dcpg_filter_motifs.py` to read and join output files.

    Paramters
    ---------
    filter_stats_file: str
        Path of stats file created with `dcpg_filter_motifs.py`.
    tomtom_file: str
        Path of Tomtom output file.
    meme_motifs: :class:`pandas.DataFrame`
        :class:`pandas.DataFrame` from `read_meme_db`.

    Returns
    -------
    :class:`pandas.DataFrame`
        :class:`pandas.DataFrame` with columns from Tomtom and statistic file.
    """
    filter_stats = pd.read_table(filter_stats_file)
    tomtom = read_tomtom(tomtom_file)
    tomtom = tomtom.sort_values(['idx', 'q-value', 'e-value'])
    tomtom = tomtom.loc[:, ~tomtom.columns.isin(['query id', 'optimal offset'])]
    d = pd.merge(filter_stats, tomtom, on='idx', how='outer')
    meme_motifs = meme_motifs.rename(columns={'id': 'target id'})
    d = pd.merge(d, meme_motifs, on='target id', how='left')
    d.index.name = None
    return d