comparison relabel_fasta.py @ 5:bbfc9638ba84 draft

First version with (partial) bioconda deps.

5:bbfc9638ba84

#!/usr/bin/env python

"""
Replace FASTA labels with new labels <PREFIX>1, <PREFIX>2,
<PREFIX>3 etc (where <PREFIX> is a user-provided argument).

Can be used to label OTUs as OTU_1, OTU_2 etc.

This is a reimplementation of the fasta_number.py script from
https://drive5.com/python/fasta_number_py.html
"""

import argparse

def relabel_fasta(fp,prefix,include_size=False):
    """
    """
    # Iterate over lines in file
    nlabel = 0
    for line in fp:
        # Strip trailing newlines
        line = line.rstrip('\n')
        if not line:
            # Skip blank lines
            continue
        elif line.startswith('>'):
            # 
            nlabel += 1
            label = line[1:].strip()
            if args.needsize:
                # Extract size from the label
                try:
                    size = filter(
                        lambda x: x.startswith("size="),
                        label.split(';'))[0]
                except Exception as ex:
                    raise Exception("Couldn't locate 'size' in "
                                    "label: %s" % label)
                yield ">%s%d;%s" % (args.prefix,
                                    nlabel,
                                    size)
            else:
                yield ">%s%d" % (args.prefix,
                                 nlabel)
        else:
            # Echo the line to output
            yield line

if __name__ == "__main__":

    # Set up command line parser
    p = argparse.ArgumentParser()
    p.add_argument("--needsize",action="store_true")
    p.add_argument("--nosize",action="store_true")
    p.add_argument("fasta")
    p.add_argument("prefix")
    
    # Process command line
    args = p.parse_args()

    # Relabel FASTA
    with open(args.fasta,'rU') as fasta:
        for line in relabel_fasta(fasta,
                                  args.prefix,
                                  include_size=args.needsize):
            print line