Merge pull request #227 from rhysnewell/fix-vamb-dastool

Fix vamb being excluded from dastool

aviary/modules/binning/scripts/das_tool.py

@@ -19,7 +19,8 @@
     binners = []
     for (binner, extension) in unrefined_binners_to_use:
         if binner not in snakemake.config['skip_binners']:
-            binners.append((f'{binner}_bins/', extension, f'data/{binner}_bins.tsv'))
+            extra = 'bins/' if binner == 'vamb' else ''
+            binners.append((f'{binner}_bins/'+extra, extension, f'data/{binner}_bins.tsv'))
 
     for (binner, extension) in refined_binners_to_use:
         if binner not in snakemake.config['skip_binners']:

test/test_integration.py

@@ -156,6 +156,42 @@ def test_short_read_recovery_fast(self):
 
         self.assertFalse(os.path.isfile(f"{output_dir}/aviary_out/data/final_contigs.fasta"))
 
+    def test_short_read_recovery_vamb(self):
+        output_dir = os.path.join("example", "test_short_read_recovery_vamb")
+        self.setup_output_dir(output_dir)
+
+        # Create inflated assembly file
+        cmd = f"cat {data}/assembly.fasta > {output_dir}/assembly.fasta"
+        multiplier = 100
+        for i in range(multiplier):
+            cmd += f" && awk '/^>/ {{print $0 \"{i}\"}} !/^>/ {{print $0}}' {data}/assembly.fasta >> {output_dir}/assembly.fasta"
+
+        subprocess.run(cmd, shell=True, check=True)
+
+        cmd = (
+            f"aviary recover "
+            f"--assembly {output_dir}/assembly.fasta "
+            f"-o {output_dir}/aviary_out "
+            f"-1 {data}/wgsim.1.fq.gz "
+            f"-2 {data}/wgsim.2.fq.gz "
+            f"--binning-only "
+            f"--skip-binners rosella semibin metabat "
+            f"--skip-qc "
+            f"--refinery-max-iterations 0 "
+            f"--conda-prefix {path_to_conda} "
+            f"-n 32 -t 32 "
+        )
+        subprocess.run(cmd, shell=True, check=True)
+
+        bin_info_path = f"{output_dir}/aviary_out/bins/bin_info.tsv"
+        self.assertTrue(os.path.isfile(bin_info_path))
+        with open(bin_info_path) as f:
+            num_lines = sum(1 for _ in f)
+        self.assertTrue(num_lines > 2)
+
+        self.assertFalse(os.path.isfile(f"{output_dir}/aviary_out/data/final_contigs.fasta"))
+
+    @unittest.skip("Skipping test due to queue submission")
     def test_short_read_recovery_queue_submission(self):
         output_dir = os.path.join("example", "test_short_read_recovery_queue_submission")
         self.setup_output_dir(output_dir)