Fix and test #197 (#8)

Mikado/subprograms/util/convert.py

@@ -20,7 +20,7 @@ def launch(args):
     current = None
     if parser.__annot_type__ == "gtf":
         out_format = "gff3"
-    elif parser.__annot_type__ in ("bed12", "gff3"):
+    elif parser.__annot_type__ in ("bed12", "gff3", "bam"):
         out_format = "gtf"
     else:
         raise TypeError("Invalid annotation type: {}".format(parser.__annot_type__))
@@ -41,7 +41,19 @@ def launch(args):
     for line in parser:
         if line.header is True:
             continue
-        if isinstance(line, BED12) or (line.is_exon is False and line.gene is None):
+        if parser.__annot_type__ == "bam":
+            # BAM file. We need to do things a little bit differently
+            if line.is_unmapped is True:
+                continue
+
+            mock_gene_counter += 1
+            gene = "gene_{mock_gene_counter}".format(**locals())
+            transcript = Transcript(line)
+            transcript.parent = gene
+            print(Gene(transcript).format(out_format, transcriptomic=args.transcriptomic), file=args.out)
+            continue
+
+        elif isinstance(line, BED12) or (line.is_exon is False and line.gene is None):
             mock_gene_counter += 1
             gene = "gene_{mock_gene_counter}".format(**locals())
             line.parent = gene
@@ -71,6 +83,7 @@ def launch(args):
             current.add(Transcript(line))
         elif line.is_exon is True:
             current.add_exon(line)
+
     if current is not None:
         print(current.format(out_format, transcriptomic=args.transcriptomic), file=args.out)
 

Mikado/tests/test_system_calls.py

@@ -35,6 +35,24 @@
 import pysam
 
 
+class ConvertCheck(unittest.TestCase):
+
+    @mark.slow
+    def test_convert_from_bam(self):
+
+        bam_inp = pkg_resources.resource_filename("Mikado.tests", "test_mRNA.bam")
+        for outp in ("gff3", "gtf", "bed12"):
+            with self.subTest(outp=outp):
+                outfile = tempfile.NamedTemporaryFile(mode="wt")
+                outfile.close()
+                sys.argv = ["", "util", "convert", "-of", outp, bam_inp, outfile.name]
+                # with self.assertRaises(SystemExit):
+                pkg_resources.load_entry_point("Mikado", "console_scripts", "mikado")()
+                self.assertGreater(os.stat(outfile.name).st_size, 0)
+                lines = [_ for _ in open(outfile.name)]
+                assert any(["TraesCS2B02G055500.1" in line for line in lines])
+
+
 # @mark.slow
 class PrepareCheck(unittest.TestCase):