Merge pull request #19 from dev/validate-num-columns-in-refgene-file

Validate number of columns in refgene file
counsyl · Aug 20, 2015 · 5032f9a · 5032f9a
2 parents def8ab6 + 3e44f32
commit 5032f9a
Show file tree

Hide file tree

Showing 3 changed files with 10 additions and 4 deletions.
diff --git a/pyhgvs/__init__.py b/pyhgvs/__init__.py
@@ -534,7 +534,7 @@ def cdna_to_genomic_coord(transcript, coord):
     # 5' flanking sequence.
     if pos < 1:
         if transcript_strand:
-            return transcript.tx_position.chrom_start + pos - 1
+            return transcript.tx_position.chrom_start + pos
         else:
             return transcript.tx_position.chrom_stop - pos + 1
 
@@ -1358,7 +1358,7 @@ def parse_hgvs_name(hgvs_name, genome, transcript=None,
     if transcript and hgvs.transcript in genome:
         # Reference sequence is directly known, use it.
         genome = GenomeSubset(genome, transcript.tx_position.chrom,
-                              transcript.tx_position.chrom_start - 1,
+                              transcript.tx_position.chrom_start,
                               transcript.tx_position.chrom_stop,
                               hgvs.transcript)
 

diff --git a/pyhgvs/utils.py b/pyhgvs/utils.py
@@ -39,22 +39,28 @@ def read_refgene(infile):
         if line.startswith('#'):
             continue
         row = line.rstrip('\n').split('\t')
+        if len(row) != 16:
+            raise ValueError(
+                'File has incorrect number of columns '
+                'in at least one line.', code='invalid')
 
         # Skip trailing ,
         exon_starts = map(int, row[9].split(',')[:-1])
         exon_ends = map(int, row[10].split(',')[:-1])
+        exon_frames = map(int, row[15].split(',')[:-1])
         exons = zip(exon_starts, exon_ends)
 
         yield {
             'chrom': row[2],
-            'start': int(row[4]) + 1,
+            'start': int(row[4]),
             'end': int(row[5]),
             'id': row[1],
             'strand': row[3],
             'cds_start': int(row[6]),
             'cds_end': int(row[7]),
             'gene_name': row[12],
             'exons': exons,
+            'exon_frames': exon_frames
         }
 
 

diff --git a/setup.py b/setup.py
@@ -20,7 +20,7 @@ def main():
 
     setup(
         name='pyhgvs',
-        version='0.9.5',
+        version='0.9.6',
         description='HGVS name parsing and formatting',
         long_description=description,
         author='Matt Rasmussen',