| 1 | #!/usr/bin/env python |
|---|
| 2 | """ |
|---|
| 3 | Input: fasta, minimal length, maximal length |
|---|
| 4 | Output: fasta |
|---|
| 5 | Return sequences whose lengths are within the range. |
|---|
| 6 | """ |
|---|
| 7 | import sys, os |
|---|
| 8 | |
|---|
| 9 | assert sys.version_info[:2] >= ( 2, 4 ) |
|---|
| 10 | |
|---|
| 11 | def stop_err( msg ): |
|---|
| 12 | sys.stderr.write( msg ) |
|---|
| 13 | sys.exit() |
|---|
| 14 | |
|---|
| 15 | def __main__(): |
|---|
| 16 | infile = sys.argv[1] |
|---|
| 17 | title_col = sys.argv[2] |
|---|
| 18 | seq_col = sys.argv[3] |
|---|
| 19 | outfile = sys.argv[4] |
|---|
| 20 | |
|---|
| 21 | if title_col == None or title_col == 'None' or seq_col == None or seq_col == 'None': |
|---|
| 22 | stop_err( "Columns not specified." ) |
|---|
| 23 | try: |
|---|
| 24 | seq_col = int( seq_col ) - 1 |
|---|
| 25 | except: |
|---|
| 26 | stop_err( "Invalid Sequence Column: %s." %str( seq_col ) ) |
|---|
| 27 | |
|---|
| 28 | title_col_list = title_col.split( ',' ) |
|---|
| 29 | out = open( outfile, 'w' ) |
|---|
| 30 | skipped_lines = 0 |
|---|
| 31 | first_invalid_line = 0 |
|---|
| 32 | invalid_line = "" |
|---|
| 33 | i = 0 |
|---|
| 34 | |
|---|
| 35 | for i, line in enumerate( open( infile ) ): |
|---|
| 36 | error = False |
|---|
| 37 | line = line.rstrip( '\r\n' ) |
|---|
| 38 | if line and not line.startswith( '#' ): |
|---|
| 39 | fields = line.split( '\t' ) |
|---|
| 40 | fasta_title = [] |
|---|
| 41 | for j in title_col_list: |
|---|
| 42 | try: |
|---|
| 43 | j = int( j ) - 1 |
|---|
| 44 | fasta_title.append( fields[j] ) |
|---|
| 45 | except: |
|---|
| 46 | skipped_lines += 1 |
|---|
| 47 | if not invalid_line: |
|---|
| 48 | first_invalid_line = i + 1 |
|---|
| 49 | invalid_line = line |
|---|
| 50 | error = True |
|---|
| 51 | break |
|---|
| 52 | if not error: |
|---|
| 53 | try: |
|---|
| 54 | fasta_seq = fields[seq_col] |
|---|
| 55 | if fasta_title[0].startswith( ">" ): |
|---|
| 56 | fasta_title[0] = fasta_title[0][1:] |
|---|
| 57 | print >> out, ">%s\n%s" % ( "_".join( fasta_title ), fasta_seq ) |
|---|
| 58 | except: |
|---|
| 59 | skipped_lines += 1 |
|---|
| 60 | if not invalid_line: |
|---|
| 61 | first_invalid_line = i + 1 |
|---|
| 62 | invalid_line = line |
|---|
| 63 | out.close() |
|---|
| 64 | |
|---|
| 65 | if skipped_lines > 0: |
|---|
| 66 | print 'Data issue: skipped %d blank or invalid lines starting at #%d: "%s"' % ( skipped_lines, first_invalid_line, invalid_line ) |
|---|
| 67 | |
|---|
| 68 | if __name__ == "__main__" : __main__() |
|---|