Convert ENCODE DCC files.txt file on UCSC site to a tab-delimited file, that can be loaded by spreadsheet or other scripts.

filestxt2matrix.py

#!/usr/bin/env python
# Time-stamp: <2013-09-17 17:17:21 Tao Liu>

import os
import sys
# ------------------------------------
# Main function
# ------------------------------------
def main():
    if len(sys.argv) < 2:
        sys.stderr.write( "need 1 paras: %s <files.txt>\n" % sys.argv[ 0 ] )
        sys.exit( 1 )

    filesfhd = open( sys.argv[ 1 ], "r" )
    
    # search for all meta types
    metatypes = set( [ "filename" ] )      # a set

    for l in filesfhd:
        l = l.rstrip( )
        ( fn, meta ) = l.split( "\t" )
        meta_s = meta.split( "; " )
        for eachmeta in meta_s:
            ( metat, metav ) = eachmeta.split( "=" )
            metatypes.add( metat )

    filesfhd.seek( 0 )

    #for metat in metatypes:
    print "\t".join( metatypes )


    for l in filesfhd:
        this_dict = {}
        l = l.rstrip()
        ( fn, meta ) = l.split( "\t" )
        this_dict[ "filename" ] = fn
        meta_s = meta.split( "; " )
        for eachmeta in meta_s:
            ( metat, metav ) = eachmeta.split( "=" )
            this_dict[ metat ] = metav

        # print
        to_write = []
        for metat in metatypes:
            if this_dict.has_key( metat ):
                to_write.append( this_dict[ metat ] )
            else:
                to_write.append( '.' )
        print "\t".join( to_write )


if __name__ == '__main__':
    main()