Mercurial > repos > bebatut > format_metaphlan2_output
changeset 1:1e74cb2c8e67 draft
"planemo upload for repository https://github.com/ASaiM/galaxytools/tree/master/tools/format_metaphlan2_output/ commit 2cc71b230101205641d7fafa822d4ab3d398066a"
| author | bebatut |
|---|---|
| date | Mon, 14 Sep 2020 09:52:15 +0000 |
| parents | 2bfa9b200600 |
| children | 370b56f8a02d |
| files | format_metaphlan2_output.py format_metaphlan2_output.xml |
| diffstat | 2 files changed, 51 insertions(+), 53 deletions(-) [+] |
line wrap: on
line diff
--- a/format_metaphlan2_output.py Wed Apr 20 07:52:41 2016 -0400 +++ b/format_metaphlan2_output.py Mon Sep 14 09:52:15 2020 +0000 @@ -1,56 +1,52 @@ #!/usr/bin/env python # -*- coding: utf-8 -*- -import sys -import os import argparse -import re + -taxo_level_correspondance = {} -taxo_level_correspondance['k'] = 'kingdom' -taxo_level_correspondance['p'] = 'phylum' -taxo_level_correspondance['c'] = 'class' -taxo_level_correspondance['o'] = 'order' -taxo_level_correspondance['f'] = 'family' -taxo_level_correspondance['g'] = 'genus' -taxo_level_correspondance['s'] = 'species' -taxo_level_correspondance['t'] = 'strains' +taxo_level_corresp = { + 'k': 'kingdom', + 'p': 'phylum', + 'c': 'class', + 'o': 'order', + 'f': 'family', + 'g': 'genus', + 's': 'species', + 't': 'strains'} + def write_taxo_abundance(output_files, level, taxo, abundance): - if not taxo_level_correspondance.has_key(level): + if level not in taxo_level_corresp: raise ValueError(level + ' is not a know taxonomic level') - output_files[taxo_level_correspondance[level]].write(taxo + '\t') - output_files[taxo_level_correspondance[level]].write(abundance + '\n') + f_n = taxo_level_corresp[level] + output_files[f_n].write(taxo + '\t') + output_files[f_n].write(abundance + '\n') + def format_metaphlan2_output(args): - taxo_levels_abundance_files = {} - taxo_levels_abundance_files['kingdom'] = open(args.kingdom_abundance_file, 'w') - taxo_levels_abundance_files['phylum'] = open(args.phylum_abundance_file, 'w') - taxo_levels_abundance_files['class'] = open(args.class_abundance_file, 'w') - taxo_levels_abundance_files['order'] = open(args.order_abundance_file, 'w') - taxo_levels_abundance_files['family'] = open(args.family_abundance_file, 'w') - taxo_levels_abundance_files['genus'] = open(args.genus_abundance_file, 'w') - taxo_levels_abundance_files['species'] = open(args.species_abundance_file, 'w') - taxo_levels_abundance_files['strains'] = open(args.strains_abundance_file, 'w') - - for taxo_level_file in taxo_levels_abundance_files: - taxo_levels_abundance_files[taxo_level_file].write(taxo_level_file + '\t') - taxo_levels_abundance_files[taxo_level_file].write('abundance\n') + taxo_levels_abund_f = { + 'kingdom': open(args.kingdom_abundance_file, 'w'), + 'phylum': open(args.phylum_abundance_file, 'w'), + 'class': open(args.class_abundance_file, 'w'), + 'order': open(args.order_abundance_file, 'w'), + 'family': open(args.family_abundance_file, 'w'), + 'genus': open(args.genus_abundance_file, 'w'), + 'species': open(args.species_abundance_file, 'w'), + 'strains': open(args.strains_abundance_file, 'w') + } - with open(args.metaphlan2_output, 'r') as input_file: - with open(args.all_taxo_level_abundance_file, 'w') as output_file: - output_file.write("kingdom\t") - output_file.write("phylum\t") - output_file.write("class\t") - output_file.write("order\t") - output_file.write("family\t") - output_file.write("genus\t") - output_file.write("species\t") - output_file.write("strains\t") - output_file.write("abundance\n") + for taxo_level_f in taxo_levels_abund_f: + s = taxo_level_f + '\t' + 'abundance\n' + taxo_levels_abund_f[taxo_level_f].write(s) + + with open(args.metaphlan2_output, 'r') as input_f: + with open(args.all_taxo_level_abundance_file, 'w') as output_f: + s = "kingdom\tphylum\tclass\torder\tfamily\t" + s += "genus\tspecies\tstrains\tabundance\n" + output_f.write(s) + levels_number = 8 - - for line in input_file.readlines(): + for line in input_f.readlines(): if line.startswith("#"): continue @@ -61,23 +57,26 @@ split_taxo = all_taxo.split('|') for level in split_taxo: taxo = level.split('__')[1] - taxo = taxo.replace("_"," ") - output_file.write(taxo + '\t') + taxo = taxo.replace("_", " ") + output_f.write(taxo + '\t') for i in range(len(split_taxo), levels_number): - output_file.write('\t') + output_f.write('\t') - output_file.write(abundance + "\n") - + output_f.write(abundance + "\n") last_taxo_level = split_taxo[-1].split('__') - taxo = last_taxo_level[1].replace("_"," ") + taxo = last_taxo_level[1].replace("_", " ") level = last_taxo_level[0] - write_taxo_abundance(taxo_levels_abundance_files, level, taxo, + write_taxo_abundance( + taxo_levels_abund_f, + level, + taxo, abundance) - for taxo_level_file in taxo_levels_abundance_files: - taxo_levels_abundance_files[taxo_level_file].close() + for taxo_level_f in taxo_levels_abund_f: + taxo_levels_abund_f[taxo_level_f].close() + if __name__ == '__main__': parser = argparse.ArgumentParser() @@ -91,7 +90,6 @@ parser.add_argument('--genus_abundance_file', required=True) parser.add_argument('--species_abundance_file', required=True) parser.add_argument('--strains_abundance_file', required=True) - args = parser.parse_args() - format_metaphlan2_output(args) \ No newline at end of file + format_metaphlan2_output(args)
--- a/format_metaphlan2_output.xml Wed Apr 20 07:52:41 2016 -0400 +++ b/format_metaphlan2_output.xml Mon Sep 14 09:52:15 2020 +0000 @@ -1,4 +1,4 @@ -<tool id="format_metaphlan2_output" name="Format MetaPhlAn2" version="0.1.0"> +<tool id="format_metaphlan2_output" name="Format MetaPhlAn2" version="0.2.0"> <description>output to extract abundance at different taxonomic levels</description> <requirements>