Removing redundancies in a loop using python

Question

I have a huge input file that looks like this,

contig  protein start end
con1    P1  140 602
con1    P2  140 602
con1    P3  232 548
con2    P4  335 801
con2    P5  642 732
con2    P6  335 779
con2    P7  729 812
con3    P8  17  348
con3    P9  16  348

I would like to remove homologous P's or redundant P's, which I assume are those that have same start and end sites and the ones that have smaller start or end sites, respectively. So my output files will be like this, file.txt

con1    P1  140 602
con1    P3  232 548
con2    P4  335 801
con2    P7  729 812

Attempted script, for some reason it doesn't meet both conditions,

from itertools import groupby
def non_homolog(hits):
    nonhomolog=[]
    overst = False
    for i in range(1,len(hits)):
        (p, c) = hits[i-1], hits[i]
        if p[2] <= c[2] and c[3] <= p[3]:
            if not overst: nonhomolog.append(c)
            nonhomolog.append(c)
            overst = True   
    return nonhomolog

fh = open('example.txt')
oh = open('nonhomologs.txt', 'w')
for qid, grp in groupby(fh, lambda l: l.split()[0]):
    hits = []
    for line in grp:
        hsp = line.split()
        hsp[2], hsp[3] = int(hsp[2]), int(hsp[3])
        hits.append(hsp)
    hits.sort(key=lambda x: x[2])
    if non_homolog(hits):
        for hit in hits:
            oh.write('\t'.join([str(f) for f in hit])+'\n')

Answer 1

Try this on for size:

# this code assumes Python 2.7
from itertools import groupby, izip
from operator import attrgetter

INPUT    = "file.txt"
HOMO_YES = "homologs.txt"
HOMO_NO  = "nonhomologs.txt"
MAX_DIFF = 5

class Row:
    __slots__ = ["line", "con", "protein", "start", "end"]

    def __init__(self, s):
        self.line    = s.rstrip()
        data         = s.split()
        self.con     = data[0]
        self.protein = data[1]
        self.start   = int(data[2])
        self.end     = int(data[3])

    def __str__(self):
        return self.line

def count_homologs(items, max_diff=MAX_DIFF):
    num_items  = len(items)
    counts     = [0] * num_items
    # first item
    for i, item_i in enumerate(items):
        max_start = item_i.start + max_diff
        max_end   = item_i.end   + max_diff
        # second item
        for j in xrange(i+1, num_items):
            item_j = items[j]
            if item_j.start > max_start:
                break
            elif item_j.end <= max_end:
                counts[i] += 1
                counts[j] += 1
    return counts

def main():
    with open(INPUT) as inf, open(HOMO_YES, "w") as outhomo, open(HOMO_NO, "w") as outnothomo:
        # skip header
        next(inf, '')
        rows = (Row(line) for line in inf)

        for con, item_iter in groupby(rows, key=attrgetter("con")):
            # per-con list of Rows sorted by start,end
            items = sorted(item_iter, key=attrgetter("start", "end"))
            # get #homologs for each item
            counts = count_homologs(items)
            # do output
            for c,item in izip(counts, items):
                if c:
                    outhomo.write(str(item) + "\n")
                else:
                    outnothomo.write(str(item) + "\n")

if __name__=="__main__":
    main()

on your given data, produces:

=== homologs.txt ===

con1    P1  140 602
con1    P2  140 602
con3    P9  16  348
con3    P8  17  348

=== nonhomologs.txt ===

con1    P3  232 548
con2    P6  335 779
con2    P4  335 801
con2    P5  642 732
con2    P7  729 812