Merge pull request #67 from jlanga/devel

Devel

Author: jlanga

exfi/__init__.py

@@ -3,4 +3,4 @@
 filters.
 """
 
-__version__ = '1.5.4'
+__version__ = '1.5.5'

exfi/arguments.py

@@ -328,8 +328,8 @@ def add_gff3_type(parser):
     parser.add_argument(
         '--type-gff3', '-m',
         type=str,
-        help='Source of the GFF3 type (ensembl or gmap)',
-        choices=['ensembl', 'gmap'],
+        help='Source of the GFF3 type (ensembl, gmap, ncbi)',
+        choices=['ensembl', 'gmap', 'ncbi'],
         default='ensembl',
         metavar='STR',
         dest='gff3_type'

exfi/io/gff3_to_bed.py

@@ -49,6 +49,12 @@ def gff3_to_bed3(gff3_in, mode="ensembl"):
         exons['transcript_id'] = exons['attributes']\
             .str.split(";").str[1]\
             .str.extract(r'Name=([\w\d.-_]+)')
+    elif mode == "ncbi":
+        logging.info("gff3 comes from NCBI Genomes")
+        exons = raw[raw['type'] == 'exon'].drop(columns='type')
+        exons['transcript_id'] = exons.attributes\
+            .str.extract(r"transcript_id=([A-Za-z0-9_.]+)")
+        exons = exons.dropna()
     else:
         logging.info('gff3 comes from ensembl')
         exons = raw[raw['type'] == 'exon'].drop(columns='type')