annotate format_taxon_list.py @ 2:d0d9e52a46b8 draft default tip

planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/checkm commit 3ae2bd72f789518e95ef65b97e0e5ac90165e113
author iuc
date Mon, 02 Sep 2024 13:51:24 +0000
parents 878d742dacf0
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
878d742dacf0 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/checkm commit 2a3b068a98bf0e913dc03e0d5c2182cfd102cf27
iuc
parents:
diff changeset
1 from pathlib import Path
878d742dacf0 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/checkm commit 2a3b068a98bf0e913dc03e0d5c2182cfd102cf27
iuc
parents:
diff changeset
2
878d742dacf0 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/checkm commit 2a3b068a98bf0e913dc03e0d5c2182cfd102cf27
iuc
parents:
diff changeset
3 from checkm.taxonParser import TaxonParser
878d742dacf0 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/checkm commit 2a3b068a98bf0e913dc03e0d5c2182cfd102cf27
iuc
parents:
diff changeset
4 from checkm.util.taxonomyUtils import taxonomicRanks
878d742dacf0 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/checkm commit 2a3b068a98bf0e913dc03e0d5c2182cfd102cf27
iuc
parents:
diff changeset
5
878d742dacf0 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/checkm commit 2a3b068a98bf0e913dc03e0d5c2182cfd102cf27
iuc
parents:
diff changeset
6 if __name__ == '__main__':
878d742dacf0 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/checkm commit 2a3b068a98bf0e913dc03e0d5c2182cfd102cf27
iuc
parents:
diff changeset
7 tool_data_dp = Path('tool-data')
878d742dacf0 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/checkm commit 2a3b068a98bf0e913dc03e0d5c2182cfd102cf27
iuc
parents:
diff changeset
8
878d742dacf0 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/checkm commit 2a3b068a98bf0e913dc03e0d5c2182cfd102cf27
iuc
parents:
diff changeset
9 # get all available marker sets
878d742dacf0 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/checkm commit 2a3b068a98bf0e913dc03e0d5c2182cfd102cf27
iuc
parents:
diff changeset
10 taxonParser = TaxonParser()
878d742dacf0 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/checkm commit 2a3b068a98bf0e913dc03e0d5c2182cfd102cf27
iuc
parents:
diff changeset
11 taxonMarkerSets = taxonParser.readMarkerSets()
878d742dacf0 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/checkm commit 2a3b068a98bf0e913dc03e0d5c2182cfd102cf27
iuc
parents:
diff changeset
12
878d742dacf0 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/checkm commit 2a3b068a98bf0e913dc03e0d5c2182cfd102cf27
iuc
parents:
diff changeset
13 # create a table per rank
878d742dacf0 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/checkm commit 2a3b068a98bf0e913dc03e0d5c2182cfd102cf27
iuc
parents:
diff changeset
14 for rank in taxonomicRanks:
878d742dacf0 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/checkm commit 2a3b068a98bf0e913dc03e0d5c2182cfd102cf27
iuc
parents:
diff changeset
15 rank_fp = tool_data_dp / Path("%s.loc.sample" % rank)
878d742dacf0 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/checkm commit 2a3b068a98bf0e913dc03e0d5c2182cfd102cf27
iuc
parents:
diff changeset
16 with rank_fp.open('w') as rank_f:
878d742dacf0 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/checkm commit 2a3b068a98bf0e913dc03e0d5c2182cfd102cf27
iuc
parents:
diff changeset
17 rank_f.write('# File generated by format_taxon_list.py script\n')
878d742dacf0 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/checkm commit 2a3b068a98bf0e913dc03e0d5c2182cfd102cf27
iuc
parents:
diff changeset
18 rank_f.write('# taxon\tdescription\n')
878d742dacf0 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/checkm commit 2a3b068a98bf0e913dc03e0d5c2182cfd102cf27
iuc
parents:
diff changeset
19 for taxon in sorted(taxonMarkerSets[rank]):
878d742dacf0 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/checkm commit 2a3b068a98bf0e913dc03e0d5c2182cfd102cf27
iuc
parents:
diff changeset
20 markerSet = taxonMarkerSets[rank][taxon]
878d742dacf0 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/checkm commit 2a3b068a98bf0e913dc03e0d5c2182cfd102cf27
iuc
parents:
diff changeset
21 numMarkers, numMarkerSets = markerSet.size()
878d742dacf0 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/checkm commit 2a3b068a98bf0e913dc03e0d5c2182cfd102cf27
iuc
parents:
diff changeset
22 rank_f.write('{taxon}\t{taxon} ({markerSet.numGenomes} genomes, {numMarkers} marker genes, {numMarkerSets} marker sets)\n')