summarise: Add --num-decimal-places.

singlem/condense.py

@@ -913,6 +913,8 @@ def each_sample_wise(io):
 
         for row in reader:
             (sample, coverage, taxonomy) = row
+            if float(coverage) == 0:
+                continue
             if sample != current_sample:
                 if current_sample is not None:
                     yield CondensedCommunityProfile(current_sample, current_root)

singlem/main.py

@@ -357,7 +357,8 @@ def add_less_common_pipe_arguments(argument_group):
     summarise_taxonomic_profile_output_args.add_argument('--output-species-by-site-level', help="Output site by species level to this file. Requires --output-species-by-site-relative-abundance.", choices=['species','genus','family','order','class','phylum','domain'], default='species')
     summarise_taxonomic_profile_output_args.add_argument('--output-species-by-site-relative-abundance-prefix', metavar='PATH_PREFIX', help="Output site by species relative abundance to this file prefix. One file will be written for each taxonomic level.")
     summarise_taxonomic_profile_output_args.add_argument('--output-filled-taxonomic-profile', metavar='FILE', 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.")
-    summarise_taxonomic_profile_output_args.add_argument('--output-taxonomic-profile-with-extras', metavar='FILE', help="Output a taxonomic profile with extra information (coverage, 'filled' coverage, relative abundance, taxonomy level). ")
+    summarise_taxonomic_profile_output_args.add_argument('--output-taxonomic-profile-with-extras', metavar='FILE', help="Output a taxonomic profile with extra information (coverage, 'filled' coverage, relative abundance, taxonomy level).")
+    summarise_taxonomic_profile_output_args.add_argument('--num-decimal-places', metavar='INT', type=int, help="Number of decimal places to report in the coverage column of the --output-taxonomic-profile-with-extras [default: 2].")
     summarise_taxonomic_profile_output_args.add_argument('--output-taxonomic-level-coverage', metavar='FILE', help="Output summary of how much coverage has been assigned to each taxonomic level in a taxonomic profile to a TSV file.")
 
     summarise_otu_table_input_args = summarise_parser.add_argument_group('OTU table input')
@@ -844,6 +845,8 @@ def get_min_taxon_coverage(args, subparser='pipe'):
             raise Exception("--output-filled-taxonomic-profile requires --input-taxonomic-profiles to be defined")
         if args.output_taxonomic_profile_with_extras and not args.input_taxonomic_profiles:
             raise Exception("--output-taxonomic-profile-with-extras requires --input-taxonomic-profiles to be defined")
+        if args.num_decimal_places and not args.output_taxonomic_profile_with_extras:
+            raise Exception("--num-decimal-places currently requires --output-taxonomic-profile-with-extras to be defined")
 
         if args.stream_inputs or args.unaligned_sequences_dump_file:
             from singlem.otu_table_collection import StreamingOtuTableCollection
@@ -1034,7 +1037,8 @@ def get_min_taxon_coverage(args, subparser='pipe'):
                 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)
+                        output_taxonomic_profile_extras_io = f,
+                        num_decimal_places = args.num_decimal_places)
             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

@@ -617,8 +617,11 @@ def write_filled_taxonomic_profile(**kwargs):
     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')
+        num_decimal_places = kwargs.pop('num_decimal_places') # Default to 2 below
         if len(kwargs) > 0:
             raise Exception("Unexpected arguments detected: %s" % kwargs)
+        if num_decimal_places is None:
+            num_decimal_places = 2
 
         logging.info("Writing taxonomic profile with extras")
 
@@ -642,9 +645,9 @@ def write_taxonomic_profile_with_extras(**kwargs):
                         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(round(wn.coverage, num_decimal_places)),
+                            str(round(full_coverage, num_decimal_places)),
+                            str(round(full_coverage / total_coverage * 100, num_decimal_places)),
                             str(levels[level]),
                             '; '.join(wn.get_taxonomy())
                         ]), file=output_io)