add cumulative_coassemblies output to record previous coassemblies

loaded automatically by iterate

Author: AroneyS

ibis/ibis.py

@@ -728,10 +728,25 @@ def iterate(args):
             }
 
     if args.coassemble_output:
+        logging.info("Processing previous Ibis coassemble run")
         coassemble_appraise_dir = os.path.join(os.path.abspath(args.coassemble_output), "appraise")
         args.coassemble_unbinned = os.path.join(coassemble_appraise_dir, "unbinned.otu_table.tsv")
         args.coassemble_binned = os.path.join(coassemble_appraise_dir, "binned.otu_table.tsv")
 
+        if args.exclude_coassemblies:
+            new_exclude_coassemblies = args.exclude_coassemblies
+        else:
+            new_exclude_coassemblies = []
+
+        cumulative_path = os.path.join(args.coassemble_output, "target", "cumulative_coassemblies.tsv")
+        if os.path.isfile(cumulative_path):
+            with open(cumulative_path) as f:
+                cumulative_exclude = f.read().splitlines()
+        else:
+            cumulative_exclude = []
+
+        args.exclude_coassemblies = cumulative_exclude + new_exclude_coassemblies
+
     try:
         args.coassemble_unbinned
     except AttributeError:
@@ -774,6 +789,14 @@ def iterate(args):
 
     coassemble(args)
 
+    elusive_clusters_path = os.path.join(args.output, "coassemble", "target", "elusive_clusters.tsv")
+    if os.path.isfile(elusive_clusters_path):
+        elusive_clusters = pl.read_csv(elusive_clusters_path, separator="\t")
+        new_coassemblies = elusive_clusters.get_column("samples").to_list()
+
+        with open(os.path.join(args.output, "coassemble", "target", "cumulative_coassemblies.tsv"), "w") as f:
+            f.write("\n".join(args.exclude_coassemblies + new_coassemblies) + "\n")
+
     if args.elusive_clusters and not args.dryrun:
         new_cluster = pl.read_csv(os.path.join(args.output, "coassemble", "target", "elusive_clusters.tsv"), separator="\t")
         for cluster in args.elusive_clusters:
@@ -785,7 +808,8 @@ def iterate(args):
                     pl.col("coassembly").map_elements(lambda x: logging.warn(f"{x} has been previously suggested"))
                     )
     elif not args.exclude_coassemblies:
-        logging.warn("Suggested coassemblies may match those from previous iterations. To check, use `--elusive-clusters`. To exclude manually, use `--exclude-coassembles`")
+        logging.warn("Suggested coassemblies may match those from previous iterations. To check, use `--elusive-clusters`.")
+        logging.warn("To exclude, provide previous run with `--coassemble-output` or use `--exclude-coassembles`.")
 
     logging.info(f"Ibis iterate complete.")
     logging.info(f"Cluster summary at {os.path.join(args.output, 'coassemble', 'summary.tsv')}")

test/data/mock_coassemble/coassemble/target/cumulative_coassemblies.tsv

@@ -0,0 +1 @@
+sample_0,sample_1

test/test_iterate.py

@@ -148,7 +148,6 @@ def test_iterate_minimal(self):
             cmd = (
                 f"ibis iterate "
                 f"--coassemble-output {MOCK_COASSEMBLE} "
-                f"--elusive-clusters {ELUSIVE_CLUSTERS} "
                 f"--singlem-metapackage {METAPACKAGE} "
                 f"--output test "
                 f"--conda-prefix {path_to_conda} "
@@ -183,6 +182,9 @@ def test_iterate_minimal(self):
                 }
             self.assertEqual(reads_2, config["reads_2"])
 
+            exclude_coassemblies = ["sample_0,sample_1"]
+            self.assertEqual(exclude_coassemblies, config["exclude_coassemblies"])
+
             cluster_path = os.path.join("test", "coassemble", "target", "elusive_clusters.tsv")
             self.assertTrue(os.path.exists(cluster_path))
             expected = "\n".join(
@@ -209,6 +211,12 @@ def test_iterate_minimal(self):
             with open(cluster_path) as f:
                 self.assertEqual(expected, f.read())
 
+            cumulative_coassemblies_path = os.path.join("test", "coassemble", "target", "cumulative_coassemblies.tsv")
+            self.assertTrue(os.path.exists(cumulative_coassemblies_path))
+            expected = "\n".join(["sample_0,sample_1", "sample_1,sample_3", ""])
+            with open(cumulative_coassemblies_path) as f:
+                self.assertEqual(expected, f.read())
+
     def test_iterate_genome_input(self):
         with in_tempdir():
             cmd = (
@@ -336,12 +344,9 @@ def test_iterate_default_config(self):
         with in_tempdir():
             cmd = (
                 f"ibis iterate "
-                f"--aviary-outputs {MOCK_COASSEMBLIES} "
-                f"--forward {SAMPLE_READS_FORWARD} "
-                f"--reverse {SAMPLE_READS_REVERSE} "
-                f"--genomes {GENOMES} "
-                f"--genome-transcripts {GENOME_TRANSCRIPTS} "
+                f"--coassemble-output {MOCK_COASSEMBLE} "
                 f"--singlem-metapackage {METAPACKAGE} "
+                f"--exclude-coassemblies sample_4,sample_5 "
                 f"--output test "
                 f"--conda-prefix {path_to_conda} "
                 f"--dryrun "
@@ -357,6 +362,7 @@ def test_iterate_default_config(self):
             self.assertEqual(config["assemble_unmapped"], False)
             self.assertEqual(config["aviary_threads"], 64)
             self.assertEqual(config["aviary_memory"], 500)
+            self.assertEqual(config["exclude_coassemblies"], ["sample_0,sample_1", "sample_4,sample_5"])
 
             self.assertTrue("Evaluating bins using CheckM2 with completeness >= 70 and contamination <= 10" in output)