source: liacs/dbdm/dbdm_4/reading_frames.py@ 61

#!/usr/bin/env python
#
# Parse FASTA file and print statistics on reading frames
# BSDLicence
# Rick van der Zwet - 0433373 - <info@rickvaderzwet.nl>
import sys

def _frame_stat(seq, start=0):
    pdict = {}
    for n in range(start,len(seq),2):
        codon = seq[n:n+3]
        if len(codon) < 3:
            continue
        if not pdict.has_key(codon):
            pdict[codon] = 1
        else:
            pdict[codon] += 1

    return(pdict)

def reading_frames(seq):
    '''Parse from left to right and right to left, at position 1,2,3 in
    in the so called nucleotide triplets 
    See: http://en.wikipedia.org/wiki/Genetic_code'''

    # Populate all empty
    final = {}
    for a in 'acgtn':
        for b in 'acgtn':
            for c in 'acgtn': 
                final[a + b + c] = [0,0,0,0,0,0]

    for start in [0,1,2]:
        print "Normal; start %i" % (start)
        retval = _frame_stat(seq,start)
        for codon,v in retval.iteritems():
            final[codon][start] += v
        print "Reverse; start %i" % (start)
        retval = _frame_stat(seq[::-1],start)
        for codon,v in retval.iteritems():
            final[codon][start+3] += v

    print "CODON :  N:0   , N:1  , N:2  ,  R:0 , R:1  , R:2  "

    for codon in sorted(final.keys()):
        print codon,"  : ", ",".join(["%6i" % x for x in final[codon]])



if __name__ == "__main__":
    # Load data
    try:
        handle = open(sys.argv[1],"rU")
    except IndexError:
        print "Usage %s <data_file>" % (sys.argv[0])
        sys.exit(64)
    except IOError:
        print "Unable to open '%s'" % (sys.argv[1])
        sys.exit(64)
    
    seq = handle.read()
    handle.close()
    reading_frames(seq)