summarise: Add --output-taxonomic-profile-with-extras.

bin/singlem

@@ -366,6 +366,7 @@ if __name__ == '__main__':
     summarise_output_args.add_argument('--output-species-by-site-level', help="Output site by species level to this file", choices=['species','genus','family','order','class','phylum','domain'], default='species')
     summarise_output_args.add_argument('--output-taxonomic-level-coverage', help="Output summary of how much coverage has been assigned to each taxonomic level in a taxonomic profile to a TSV file")
     summarise_output_args.add_argument('--output-filled-taxonomic-profile', help="Output a taxonomic profile where the coverage of each taxon includes the coverage of each of its descendent taxons e.g. the d__Bacteria entry includes the p__Patescibacteria entry. Requires --input-taxonomic-profiles.")
+    summarise_output_args.add_argument('--output-taxonomic-profile-with-extras', help="Output a taxonomic profile with extra information (coverage, 'filled' coverage, relative abundance, taxonomy level). Requires --input-taxonomic-profiles.")
 
     read_fraction_description = 'Estimate the fraction of reads from a metagenome that are assigned to Bacteria and Archaea compared to e.g. eukaryote or phage.'
     read_fraction_parser = bird_argparser.new_subparser('microbial_fraction', read_fraction_description, parser_group='Tools')
@@ -776,6 +777,7 @@ if __name__ == '__main__':
         if args.output_species_by_site_relative_abundance: num_output_types += 1
         if args.output_taxonomic_level_coverage: num_output_types += 1
         if args.output_filled_taxonomic_profile: num_output_types += 1
+        if args.output_taxonomic_profile_with_extras: num_output_types += 1
         if num_output_types != 1:
             raise Exception("Exactly 1 output type must be specified, sorry, %i were provided" % num_output_types)
         if not args.input_otu_tables and \
@@ -982,6 +984,11 @@ if __name__ == '__main__':
                     Summariser.write_filled_taxonomic_profile(
                         input_taxonomic_profile_files = args.input_taxonomic_profiles,
                         output_filled_taxonomic_profile_io = f)
+            elif args.output_taxonomic_profile_with_extras:
+                with open(args.output_taxonomic_profile_with_extras, 'w') as f:
+                    Summariser.write_taxonomic_profile_with_extras(
+                        input_taxonomic_profile_files = args.input_taxonomic_profiles,
+                        output_taxonomic_profile_extras_io = f)
             else:
                 raise Exception("Expected --output-taxonomic-profile-krona or --output-site-by-species-relative-abundance or --output-taxonomic-level-coverage to be defined, since --input-taxonomic-profiles was defined")
 

singlem/summariser.py

@@ -586,4 +586,39 @@ def write_filled_taxonomic_profile(**kwargs):
                             file=output_io)
 
         logging.info("Wrote {} filled taxonomic profiles".format(
-            len(input_taxonomic_profile_files)))
+            len(input_taxonomic_profile_files)))
+
+    @staticmethod
+    def write_taxonomic_profile_with_extras(**kwargs):
+        input_taxonomic_profile_files = kwargs.pop('input_taxonomic_profile_files')
+        output_io = kwargs.pop('output_taxonomic_profile_extras_io')
+        if len(kwargs) > 0:
+            raise Exception("Unexpected arguments detected: %s" % kwargs)
+
+        logging.info("Writing taxonomic profile with extras")
+
+        print("\t".join(["sample", "coverage", "full_coverage", "relative_abundance", "level", "taxonomy"]), file=output_io)
+
+        # For each profile
+        num_printed = 0
+        for profile_file in input_taxonomic_profile_files:
+            with open(profile_file) as f:
+                for profile in CondensedCommunityProfile.each_sample_wise(f):
+                    # First get the total coverage of each taxonomic level, to act as the numerator
+                    total_coverage = profile.tree.get_full_coverage()
+
+                    # Now write out the profile
+                    for wn in profile.breadth_first_iter():
+                        level = wn.calculate_level()
+                        full_coverage = wn.get_full_coverage()
+                        print("\t".join([
+                            profile.sample,
+                            str(wn.coverage),
+                            str(round(full_coverage, 2)),
+                            str(round(full_coverage / total_coverage * 100, 2)),
+                            str(level),
+                            '; '.join(wn.get_taxonomy())
+                        ]), file=output_io)
+                        num_printed += 1
+
+        logging.info("Wrote {} lines of taxonomic profile with extras".format(num_printed))

test/test_summariser.py

@@ -379,6 +379,25 @@ def test_fill_condensed(self):
 land       2.17    Root; d__Bacteria; p__Proteobacteria
 """)
         self.assertEqual(expected, stdout)
+
+    def test_output_taxonomic_profile_with_extras(self):
+        cmd = f'bin/singlem summarise --input-taxonomic-profile <(head -5 {path_to_data}/read_fraction/marine0.profile) <(head -5 {path_to_data}/read_fraction/marine0.profile |sed s/marine0.1/land/) '\
+            '--output-taxonomic-profile-with-extras /dev/stdout'
+        stdout = extern.run(cmd)
+        expected = re.compile(r'  +').sub('\t', """sample     coverage  full_coverage  relative_abundance  level  taxonomy
+marine0.1  0         7.17           100.0               0      Root
+marine0.1  3.64      4.2            58.58               1      Root; d__Archaea
+marine0.1  0         2.97           41.42               1      Root; d__Bacteria
+marine0.1  0.56      0.56           7.81                2      Root; d__Archaea; p__Thermoproteota
+marine0.1  0.8       0.8            11.16               2      Root; d__Bacteria; p__Desulfobacterota
+marine0.1  2.17      2.17           30.26               2      Root; d__Bacteria; p__Proteobacteria
+land       0         7.17           100.0               0      Root
+land       3.64      4.2            58.58               1      Root; d__Archaea
+land       0         2.97           41.42               1      Root; d__Bacteria
+land       0.56      0.56           7.81                2      Root; d__Archaea; p__Thermoproteota
+land       0.8       0.8            11.16               2      Root; d__Bacteria; p__Desulfobacterota
+land       2.17      2.17           30.26               2      Root; d__Bacteria; p__Proteobacteria
+""")
 
 
 if __name__ == "__main__":