Added task fastaq_to_orfs_gff

fastaq/tasks.py

@@ -190,6 +190,23 @@ def fastaq_to_mira_xml(infile, outfile):
     utils.close(fout)
 
 
+def fastaq_to_orfs_gff(infile, outfile, min_length=300):
+    seq_reader = sequences.file_reader(infile)
+    fout = utils.open_file_write(outfile)
+    for seq in seq_reader:
+        print(seq.id)
+        orfs = seq.all_orfs(min_length=min_length)
+        for coords, revcomp in orfs:
+            if revcomp:
+                strand = '-'
+            else:
+                strand = '+'
+
+            print(seq.id, 'fastaq', 'CDS', coords.start+1, coords.end+1, '.', strand, '.', sep='\t', file=fout)
+
+    utils.close(fout)
+
+
 def file_to_dict(infile, d):
     seq_reader = sequences.file_reader(infile)
     for seq in seq_reader:

fastaq/tests/data/sequences_test_orfs.gff

@@ -0,0 +1,15 @@
+1	fastaq	CDS	28	222	.	+	.
+1	fastaq	CDS	45	227	.	+	.
+1	fastaq	CDS	49	171	.	-	.
+1	fastaq	CDS	110	241	.	+	.
+1	fastaq	CDS	144	266	.	-	.
+1	fastaq	CDS	228	422	.	+	.
+1	fastaq	CDS	278	433	.	-	.
+1	fastaq	CDS	287	478	.	+	.
+1	fastaq	CDS	289	519	.	-	.
+1	fastaq	CDS	563	703	.	+	.
+1	fastaq	CDS	601	759	.	+	.
+1	fastaq	CDS	606	818	.	+	.
+1	fastaq	CDS	819	938	.	+	.
+1	fastaq	CDS	836	988	.	+	.
+1	fastaq	CDS	865	999	.	+	.

fastaq/tests/tasks_test.py

@@ -88,6 +88,15 @@ def test_fastaq_to_mira_xml(self):
         os.unlink(tmp)
 
 
+class TestFastaqToOrfsGFF(unittest.TestCase):
+    def test_fastaq_to_orfs_gff(self):
+        '''Test fastaq_to_orfs_gff'''
+        outfile = 'tmp.orfs.gff'
+        tasks.fastaq_to_orfs_gff(os.path.join(data_dir, 'sequences_test_orfs.fa'), outfile, min_length=120)
+        self.assertTrue(filecmp.cmp(os.path.join(data_dir, 'sequences_test_orfs.gff'), outfile, shallow=False))
+        os.unlink(outfile)
+
+
 class TestFilter(unittest.TestCase):
     def test_length_filter(self):
         '''Check that filtering by length works as expected'''