Fix EI-CoreBioinformatics#203

Mikado/subprograms/util/grep.py

@@ -135,17 +135,15 @@ def grep_gtf(args, gene_ids, mrna_ids):
         if not record:
             continue
         if args.genes is False:
-            if record.transcript in mrna_ids:
-                if not args.reverse:
-                    print(record, file=args.out)
-                elif args.reverse:
-                    print(record, file=args.out)
+            if record.transcript in mrna_ids and not args.reverse:
+                print(record, file=args.out)
+            elif record.transcript not in mrna_ids and args.reverse:
+                print(record, file=args.out)
         else:
-            if record.gene in gene_ids:
-                if not args.reverse:
-                    print(record, file=args.out)
-                elif args.reverse:
-                    print(record, file=args.out)
+            if record.gene in gene_ids and not args.reverse:
+                print(record, file=args.out)
+            elif record.gene not in gene_ids and args.reverse:
+                print(record, file=args.out)
 
 
 def launch(args):

Mikado/tests/test_system_calls.py

@@ -33,6 +33,7 @@
 import threading
 from time import sleep
 import pysam
+import io
 
 
 class ConvertCheck(unittest.TestCase):
@@ -1642,5 +1643,65 @@ def test_stat(self):
                 dir.cleanup()
 
 
+class GrepTest(unittest.TestCase):
+
+    @mark.slow
+    def test_grep(self):
+        files = [pkg_resources.resource_filename("Mikado.tests", fname) for fname in ["trinity.gtf", "trinity.gff3"]]
+        with io.TextIOWrapper(pkg_resources.resource_stream("Mikado.tests", "trinity.ids")) as id_file:
+            ids = [tuple(line.rstrip().split("\t")) for line in id_file]
+
+        id_temp_file = tempfile.NamedTemporaryFile("wt", suffix=".txt")
+        to_write = list(random.sample(ids, 10))
+        [print(*idline, sep="\t", file=id_temp_file) for idline in to_write]
+        id_temp_file.flush()
+
+        for fname in files:
+            with self.subTest(fname=fname):
+                form = os.path.splitext(fname)[1]
+                outfile = tempfile.NamedTemporaryFile("wt", suffix=form)
+                outfile.close()
+                self.assertFalse(os.path.exists(outfile.name))
+                sys.argv = ["mikado", "util", "grep", id_temp_file.name, fname, outfile.name]
+                pkg_resources.load_entry_point("Mikado", "console_scripts", "mikado")()
+                self.assertTrue(os.path.exists(outfile.name))
+                found = set()
+                with to_gff(outfile.name, input_format=form[1:]) as stream:
+                    for record in stream:
+                        if record.is_transcript:
+                            found.add(record.transcript)
+                self.assertEqual(len(found), 10)
+                self.assertEqual(found, set(_[0] for _ in to_write))
+
+    @mark.slow
+    def test_v_grep(self):
+        files = [pkg_resources.resource_filename("Mikado.tests", fname) for fname in ["trinity.gtf", "trinity.gff3"]]
+        with io.TextIOWrapper(pkg_resources.resource_stream("Mikado.tests", "trinity.ids")) as id_file:
+            ids = [tuple(line.rstrip().split("\t")) for line in id_file]
+
+        id_temp_file = tempfile.NamedTemporaryFile("wt", suffix=".txt")
+        to_write = list(random.sample(ids, 10))
+        others = [_ for _ in ids if _ not in to_write]
+        [print(*idline, sep="\t", file=id_temp_file) for idline in to_write]
+        id_temp_file.flush()
+
+        for fname in files:
+            with self.subTest(fname=fname):
+                form = os.path.splitext(fname)[1]
+                outfile = tempfile.NamedTemporaryFile("wt", suffix=form)
+                outfile.close()
+                self.assertFalse(os.path.exists(outfile.name))
+                sys.argv = ["mikado", "util", "grep", "-v", id_temp_file.name, fname, outfile.name]
+                pkg_resources.load_entry_point("Mikado", "console_scripts", "mikado")()
+                self.assertTrue(os.path.exists(outfile.name))
+                found = set()
+                with to_gff(outfile.name, input_format=form[1:]) as stream:
+                    for record in stream:
+                        if record.is_transcript:
+                            found.add(record.transcript)
+                self.assertEqual(len(found), len(others))
+                self.assertEqual(found, set(_[0] for _ in others))
+
+
 if __name__ == "__main__":
     unittest.main()

Mikado/tests/trinity.ids

@@ -0,0 +1,38 @@
+c73_g1_i1.mrna1.160	c73_g1_i1.path1.160
+c11_g1_i2.mrna1.111	c11_g1_i2.path1.111
+c11_g1_i1.mrna1.350	c11_g1_i1.path1.350
+c3_g1_i2.mrna1.181	c3_g1_i2.path1.181
+c3_g1_i1.mrna1.716	c3_g1_i1.path1.716
+c58_g1_i3.mrna1.19	c58_g1_i3.path1.19
+c58_g1_i2.mrna1.35	c58_g1_i2.path1.35
+c58_g1_i1.mrna1.190	c58_g1_i1.path1.190
+c58_g1_i7.mrna1.1	c58_g1_i7.path1.1
+c108_g1_i1.mrna1.104	c108_g1_i1.path1.104
+c58_g1_i11.mrna1	c58_g1_i11.path1
+c58_g1_i5.mrna1.3	c58_g1_i5.path1.3
+c58_g1_i10.mrna1	c58_g1_i10.path1
+c68_g1_i1.mrna1.173	c68_g1_i1.path1.173
+c58_g1_i11.mrna2	c58_g1_i11.path2
+c58_g1_i8.mrna2	c58_g1_i8.path2
+c58_g1_i10.mrna2	c58_g1_i10.path2
+c58_g1_i12.mrna1	c58_g1_i12.path1
+c58_g1_i9.mrna1	c58_g1_i9.path1
+c58_g1_i4.mrna1.6	c58_g1_i4.path1.6
+c58_g1_i6.mrna1.3	c58_g1_i6.path1.3
+c58_g1_i8.mrna1	c58_g1_i8.path1
+c113_g1_i1.mrna1.94	c113_g1_i1.path1.94
+c77_g1_i1.mrna1.153	c77_g1_i1.path1.153
+c109_g1_i1.mrna1.102	c109_g1_i1.path1.102
+c60_g1_i1.mrna1.189	c60_g1_i1.path1.189
+c115_g1_i1.mrna1.88	c115_g1_i1.path1.88
+c152_g1_i1.mrna1.66	c152_g1_i1.path1.66
+c21_g1_i1.mrna1.302	c21_g1_i1.path1.302
+c41_g1_i1.mrna1.224	c41_g1_i1.path1.224
+c6_g1_i1.mrna1.412	c6_g1_i1.path1.412
+c74_g1_i1.mrna1.154	c74_g1_i1.path1.154
+c71_g1_i1.mrna1.167	c71_g1_i1.path1.167
+c114_g1_i1.mrna1.89	c114_g1_i1.path1.89
+c137_g1_i1.mrna1.77	c137_g1_i1.path1.77
+c37_g1_i2.mrna1.66	c37_g1_i2.path1.66
+c37_g1_i1.mrna1.234	c37_g1_i1.path1.234
+c120_g1_i1.mrna1.89	c120_g1_i1.path1.89