Merge pull request #3 from rpetit3/rp3-add-gzip-support

Add gzip support for inputs

hicap/annotation.py

@@ -43,7 +43,7 @@ def collect_orfs(fasta_fp, model_fp):
 
 def annotate(query_fp, model_fp):
     logging.debug('Annotating %s using Prodigal', query_fp)
-    command = 'prodigal -f sco -i %s -m -t %s'
+    command = 'gzip -c -d %s | prodigal -f sco -i /dev/stdin -m -t %s' if query_fp.name.endswith('gz') else 'prodigal -f sco -i %s -m -t %s'
     result = utility.execute_command(command % (query_fp, model_fp))
     # Prodigal includes \r from FASTAs, causing problems with the output. Remove \r here
     return result.stdout.replace('\r', '')

hicap/report.py

@@ -26,7 +26,7 @@ def __init__(self):
 
 def write_outputs(locus_data, args):
     # Report
-    prefix = args.query_fp.stem
+    prefix = pathlib.Path(args.query_fp.stem).stem if args.query_fp.name.endswith('gz') else args.query_fp.stem
     output_report_fp = pathlib.Path(args.output_dir, '%s.tsv' % prefix)
     fasta = utility.read_fasta(args.query_fp)
     summary_data = create_summary(locus_data, fasta)

hicap/utility.py

@@ -82,7 +82,9 @@ def check_dependencies():
 
 
 def read_fasta(filepath):
-    with filepath.open('r') as fh:
+    import gzip
+
+    with (gzip.open(filepath, 'rt') if filepath.name.endswith('gz') else open(filepath, 'rt')) as fh:
         # TODO: cleaner way to do this?
         fasta = {desc: seq for desc, seq in Bio.SeqIO.FastaIO.SimpleFastaParser(fh)}
     if not fasta: