Adds option to keep or split reads in discordant pairs

onecodex/scripts/filter_fastq.py

@@ -16,7 +16,7 @@ def with_progress_bar(length, ix, *args, **kwargs):
     if label:
         bar_kwargs['label'] = label
     with click.progressbar(**bar_kwargs) as bar:
-        return ix(*args, **kwargs, cb=bar.update)
+        return ix(cb=bar.update, *args, **kwargs)
 
 
 def filter_rows_by_taxid(results, tax_ids, cb=None):
@@ -36,10 +36,11 @@ def get_record_count(fp):
     return counter
 
 
-def get_filtered_filename(full_filename):
-    filename, ext = os.path.splitext(full_filename.rstrip('.gz')
+def get_filtered_filename(filepath):
+    filename = os.path.split(filepath)[-1]
+    prefix, ext = os.path.splitext(filename.rstrip('.gz')
                                                   .rstrip('gzip'))
-    return '{}.filtered{}'.format(filename, ext), ext
+    return '{}.filtered{}'.format(prefix, ext), ext
 
 
 def fetch_readlevel_results(api, classification_id):
@@ -65,12 +66,16 @@ def write_fastx_record(record, handler):
 @click.command()
 @click.argument('classification')
 @click.argument('fastx', type=click.Path())
-@click.option('-r', '--reverse', type=click.Path(), help='The reverse read '
-              'file, optionally')
 @click.option('-t', '--tax-ids', required=True, help='A comma-delimited list '
               'of tax ID\'s to retain')
+@click.option('-r', '--reverse', type=click.Path(), help='The reverse (R2) '
+              'read file, optionally')
+@click.option('--split-pairs/--keep-pairs', default=False, help='Keep only '
+              'the read pair member that matches the list of tax ID\'s')
+@click.option('-o', '--out', default='.', type=click.Path(), help='Where '
+              'to put the filtered outputs')
 @click.pass_context
-def cli(ctx, classification, fastx, reverse, tax_ids):
+def cli(ctx, classification, fastx, reverse, tax_ids, split_pairs, out):
     tax_ids = tax_ids.split(',')
     if not len(tax_ids):
         raise OneCodexException('You must supply at least one tax ID')
@@ -112,8 +117,10 @@ def cli(ctx, classification, fastx, reverse, tax_ids):
             )
 
     filtered_filename = get_filtered_filename(fastx)[0]
+    filtered_filename = os.path.join(out, filtered_filename)
     if reverse:
-        rev_filtered_filename, rev_ext = get_filtered_filename(reverse)
+        rev_filtered_filename = get_filtered_filename(reverse)[0]
+        rev_filtered_filename = os.path.join(out, rev_filtered_filename)
 
     click.echo('Getting FASTX record count')
     fastx_record_count = 0
@@ -138,15 +145,25 @@ def cli(ctx, classification, fastx, reverse, tax_ids):
                 open(reverse, 'rb') as reverse_file, \
                 open(filtered_filename, 'wb') as out_file, \
                 open(rev_filtered_filename, 'wb') as rev_out_file:
-            counter = 0
-            for fwd, rev in FASTXTranslator(fastx_file, reverse_file,
-                                            validate=False):
-                if counter in filtered_rows:
-                    out_file.write(fwd)
-                counter += 1
-                if counter in filtered_rows:
-                    rev_out_file.write(rev)
-                counter += 1
+            if split_pairs:
+                counter = 0
+                for fwd, rev in FASTXTranslator(fastx_file, reverse_file,
+                                                validate=False):
+                    if counter in filtered_rows:
+                        out_file.write(fwd)
+                    if (counter + 1) in filtered_rows:
+                        rev_out_file.write(rev)
+                    counter += 2
+            else:
+                counter = 0
+                for fwd, rev in FASTXTranslator(fastx_file, reverse_file,
+                                                validate=False):
+                    if counter in filtered_rows or \
+                       (counter + 1) in filtered_rows:
+                        out_file.write(fwd)
+                        rev_out_file.write(rev)
+                    counter += 2
+
     else:
         click.echo('Filtering {}'.format(fastx))
         with open(fastx, 'rb') as fastx_file, \

tests/test_cli.py

@@ -247,7 +247,6 @@ def test_paired_files(runner, upload_mocks):
         # Check with only --forward, should fail
         args = ['--api-key', '01234567890123456789012345678901', 'upload', '--forward', f]
         result5 = runner.invoke(Cli, args)
-        print('You must specify both forward and reverse files' in result5.output)
         assert 'You must specify both forward and reverse files' in result5.output
         assert result5.exit_code != 0