Mercurial > repos > devteam > concat
annotate gops_concat.py @ 5:e24888154cb7 draft default tip
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/concat commit 58cb53d87440bd45e014fb37cddc276dc79b9d94
| author | devteam |
|---|---|
| date | Mon, 25 Jun 2018 19:47:54 -0400 |
| parents | 9a8f22b1a3ed |
| children |
| rev | line source |
|---|---|
| 0 | 1 #!/usr/bin/env python |
| 2 """ | |
| 3 Concatenate two bed files. The concatenated files are returned in the | |
| 4 same format as the first. If --sameformat is specified, then all | |
| 5 columns will be treated as the same, and all fields will be saved, | |
| 6 although the output will be trimmed to match the primary input. In | |
| 7 addition, if --sameformat is specified, missing fields will be padded | |
| 8 with a period(.). | |
| 9 | |
| 10 usage: %prog in_file_1 in_file_2 out_file | |
| 11 -1, --cols1=N,N,N,N: Columns for chrom, start, end, strand in first file | |
| 12 -2, --cols2=N,N,N,N: Columns for chrom, start, end, strand in second file | |
| 13 -s, --sameformat: All files are precisely the same format. | |
| 14 """ | |
|
4
9a8f22b1a3ed
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/concat commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689
devteam
parents:
2
diff
changeset
|
15 from __future__ import print_function |
| 0 | 16 |
|
2
23abefbed3dd
planemo upload commit 33927a87ba2eee9bf0ecdd376a66241b17b3d734
devteam
parents:
0
diff
changeset
|
17 import fileinput |
|
23abefbed3dd
planemo upload commit 33927a87ba2eee9bf0ecdd376a66241b17b3d734
devteam
parents:
0
diff
changeset
|
18 import sys |
|
4
9a8f22b1a3ed
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/concat commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689
devteam
parents:
2
diff
changeset
|
19 |
|
9a8f22b1a3ed
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/concat commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689
devteam
parents:
2
diff
changeset
|
20 from bx.cookbook import doc_optparse |
|
2
23abefbed3dd
planemo upload commit 33927a87ba2eee9bf0ecdd376a66241b17b3d734
devteam
parents:
0
diff
changeset
|
21 from bx.intervals.io import GenomicInterval, NiceReaderWrapper |
|
23abefbed3dd
planemo upload commit 33927a87ba2eee9bf0ecdd376a66241b17b3d734
devteam
parents:
0
diff
changeset
|
22 from bx.intervals.operations.concat import concat |
|
23abefbed3dd
planemo upload commit 33927a87ba2eee9bf0ecdd376a66241b17b3d734
devteam
parents:
0
diff
changeset
|
23 from bx.tabular.io import ParseError |
|
23abefbed3dd
planemo upload commit 33927a87ba2eee9bf0ecdd376a66241b17b3d734
devteam
parents:
0
diff
changeset
|
24 from galaxy.tools.util.galaxyops import fail, parse_cols_arg, skipped |
| 0 | 25 |
| 26 assert sys.version_info[:2] >= ( 2, 4 ) | |
| 27 | |
|
2
23abefbed3dd
planemo upload commit 33927a87ba2eee9bf0ecdd376a66241b17b3d734
devteam
parents:
0
diff
changeset
|
28 |
| 0 | 29 def main(): |
|
2
23abefbed3dd
planemo upload commit 33927a87ba2eee9bf0ecdd376a66241b17b3d734
devteam
parents:
0
diff
changeset
|
30 sameformat = False |
| 0 | 31 |
| 32 options, args = doc_optparse.parse( __doc__ ) | |
| 33 try: | |
| 34 chr_col_1, start_col_1, end_col_1, strand_col_1 = parse_cols_arg( options.cols1 ) | |
| 35 chr_col_2, start_col_2, end_col_2, strand_col_2 = parse_cols_arg( options.cols2 ) | |
|
2
23abefbed3dd
planemo upload commit 33927a87ba2eee9bf0ecdd376a66241b17b3d734
devteam
parents:
0
diff
changeset
|
36 if options.sameformat: |
|
23abefbed3dd
planemo upload commit 33927a87ba2eee9bf0ecdd376a66241b17b3d734
devteam
parents:
0
diff
changeset
|
37 sameformat = True |
| 0 | 38 in_file_1, in_file_2, out_fname = args |
| 39 except: | |
| 40 doc_optparse.exception() | |
| 41 | |
| 42 g1 = NiceReaderWrapper( fileinput.FileInput( in_file_1 ), | |
| 43 chrom_col=chr_col_1, | |
| 44 start_col=start_col_1, | |
| 45 end_col=end_col_1, | |
| 46 strand_col=strand_col_1, | |
| 47 fix_strand=True ) | |
| 48 | |
| 49 g2 = NiceReaderWrapper( fileinput.FileInput( in_file_2 ), | |
| 50 chrom_col=chr_col_2, | |
| 51 start_col=start_col_2, | |
| 52 end_col=end_col_2, | |
| 53 strand_col=strand_col_2, | |
| 54 fix_strand=True ) | |
| 55 | |
| 56 out_file = open( out_fname, "w" ) | |
| 57 | |
| 58 try: | |
| 59 for line in concat( [g1, g2], sameformat=sameformat ): | |
| 60 if type( line ) is GenomicInterval: | |
| 61 out_file.write( "%s\n" % "\t".join( line.fields ) ) | |
| 62 else: | |
| 63 out_file.write( "%s\n" % line ) | |
|
4
9a8f22b1a3ed
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/concat commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689
devteam
parents:
2
diff
changeset
|
64 except ParseError as exc: |
| 0 | 65 out_file.close() |
| 66 fail( "Invalid file format: %s" % str( exc ) ) | |
| 67 | |
| 68 out_file.close() | |
| 69 | |
| 70 if g1.skipped > 0: | |
|
4
9a8f22b1a3ed
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/concat commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689
devteam
parents:
2
diff
changeset
|
71 print(skipped( g1, filedesc=" of 1st dataset" )) |
| 0 | 72 if g2.skipped > 0: |
|
4
9a8f22b1a3ed
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/concat commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689
devteam
parents:
2
diff
changeset
|
73 print(skipped( g2, filedesc=" of 2nd dataset" )) |
|
9a8f22b1a3ed
planemo upload for repository https://github.com/galaxyproject/tools-devteam/tree/master/tool_collections/gops/concat commit cae3e05d02e60f595bb8b6d77a84f030e9bd1689
devteam
parents:
2
diff
changeset
|
74 |
|
2
23abefbed3dd
planemo upload commit 33927a87ba2eee9bf0ecdd376a66241b17b3d734
devteam
parents:
0
diff
changeset
|
75 |
| 0 | 76 if __name__ == "__main__": |
| 77 main() |