load samples and genomes from old config if not provided to iterate

ibis/ibis.py

@@ -678,6 +678,18 @@ def combine_genome_singlem(genome_singlem, new_genome_singlem, path):
             f.write(g.read())
 
 def iterate(args):
+    if not (args.genomes or args.forward):
+        logging.info("Loading inputs from old config")
+        config_path = os.path.join(args.coassemble_output, "config.yaml")
+        old_config = load_config(config_path)
+
+        if not args.genomes:
+            args.genomes = [v for _,v in old_config["genomes"].items()]
+            args.no_genomes = False
+        if not args.forward:
+            args.forward = [v for _,v in old_config["reads_1"].items()]
+            args.reverse = [v for _,v in old_config["reads_2"].items()]
+
     logging.info("Evaluating new bins")
     if args.new_genomes_list:
         args.new_genomes = read_list(args.new_genomes_list)
@@ -1169,8 +1181,9 @@ def base_argument_verification(args):
         if (args.forward and args.forward_list) or (args.reverse and args.reverse_list) or (args.genomes and args.genomes_list):
             raise Exception("General and list arguments are mutually exclusive")
 
-    def coassemble_argument_verification(args):
-        base_argument_verification(args)
+    def coassemble_argument_verification(args, iterate=False):
+        if not iterate:
+            base_argument_verification(args)
         if (args.sample_query or args.sample_query_list or args.sample_query_dir) and not (args.sample_singlem or args.sample_singlem_list or args.sample_singlem_dir):
             raise Exception("Input SingleM query (--sample-query) requires SingleM otu tables (--sample-singlem) for coverage")
         if args.assemble_unmapped and args.single_assembly:
@@ -1233,7 +1246,11 @@ def coassemble_output_argument_verification(args):
             raise Exception("Single assembly is incompatible with Ibis iterate")
         if not args.aviary_outputs and not (args.new_genomes or args.new_genomes_list):
             raise Exception("New genomes or aviary outputs must be provided for iteration")
-        coassemble_argument_verification(args)
+        if (args.forward and args.forward_list) or (args.reverse and args.reverse_list) or (args.genomes and args.genomes_list):
+            raise Exception("General and list arguments are mutually exclusive")
+        if not (args.genomes or args.forward) and not args.coassemble_output:
+            raise Exception("Reference genomes or forward reads must be provided if --coassemble-output not given")
+        coassemble_argument_verification(args, iterate=True)
         iterate(args)
 
     elif args.subparser_name == "build":

test/data/mock_coassemble/config.yaml

@@ -0,0 +1,47 @@
+%YAML 1.1
+---
+Ibis_version: 0.9.6
+max_threads: 8
+coassembly_samples: []
+singlem_metapackage: /mnt/hpccs01/home/aroneys/src/ibis_dev/test/data/singlem_metapackage.smpkg
+genomes:
+  GB_GCA_013286235.1: /mnt/hpccs01/home/aroneys/src/ibis_dev/test/data/GB_GCA_013286235.1.fna
+taxa_of_interest: ''
+max_coassembly_size:
+assemble_unmapped: false
+run_qc: false
+sra: false
+prodigal_meta: false
+single_assembly: false
+no_genomes: false
+new_genomes: false
+exclude_coassemblies:
+appraise_sequence_identity: 0.86
+min_coassembly_coverage: 10
+num_coassembly_samples: 2
+max_coassembly_samples: 2
+max_recovery_samples: 20
+unmapping_min_appraised: 0.1
+unmapping_max_identity: 99
+unmapping_max_alignment: 99
+aviary_speed: fast
+run_aviary: false
+aviary_gtdbtk: gtdb_release
+aviary_checkm2: CheckM2_database
+aviary_memory: 500
+aviary_threads: 64
+test: false
+mock_sra: false
+aviary_dryrun: false
+conda_prefix: /mnt/hpccs01/home/aroneys/src/ibis_dev/test/data/.conda
+tmpdir: /tmp
+reads_1:
+  sample_1: /mnt/hpccs01/home/aroneys/src/ibis_dev/test/data/sample_1.1.fq
+  sample_2: /mnt/hpccs01/home/aroneys/src/ibis_dev/test/data/sample_2.1.fq
+  sample_3: /mnt/hpccs01/home/aroneys/src/ibis_dev/test/data/sample_3.1.fq
+reads_2:
+  sample_1: /mnt/hpccs01/home/aroneys/src/ibis_dev/test/data/sample_1.2.fq
+  sample_2: /mnt/hpccs01/home/aroneys/src/ibis_dev/test/data/sample_2.2.fq
+  sample_3: /mnt/hpccs01/home/aroneys/src/ibis_dev/test/data/sample_3.2.fq
+snakemake_profile: ''
+cluster_retries: 0

test/test_iterate.py

@@ -143,6 +143,72 @@ def test_iterate(self):
             with open(cluster_path) as f:
                 self.assertEqual(expected, f.read())
 
+    def test_iterate_minimal(self):
+        with in_tempdir():
+            cmd = (
+                f"ibis iterate "
+                f"--coassemble-output {MOCK_COASSEMBLE} "
+                f"--coassemble-unbinned {MOCK_UNBINNED} "
+                f"--coassemble-binned {MOCK_BINNED} "
+                f"--aviary-outputs {MOCK_COASSEMBLIES} "
+                f"--elusive-clusters {ELUSIVE_CLUSTERS} "
+                f"--singlem-metapackage {METAPACKAGE} "
+                f"--output test "
+                f"--conda-prefix {path_to_conda} "
+            )
+            extern.run(cmd)
+
+            config_path = os.path.join("test", "config.yaml")
+            self.assertTrue(os.path.exists(config_path))
+            config = load_configfile(config_path)
+            NEW_GENOMES = " ".join([
+                os.path.join(MOCK_COASSEMBLE, "coassemble", "coassembly_0", "recover", "bins", "final_bins", "iteration_0-coassembly_0-0.fna"),
+                os.path.join(MOCK_COASSEMBLE, "coassemble", "coassembly_0", "recover", "bins", "final_bins", "iteration_0-coassembly_0-1.fna"),
+            ])
+            genomes = {
+                os.path.splitext(os.path.basename(g))[0]: g.replace(MOCK_COASSEMBLE + "/coassemble/coassembly_0/recover/bins/final_bins/", os.getcwd() + "/test/recovered_bins/")
+                for g in (GENOMES + " " + NEW_GENOMES).split(" ")
+                }
+            self.assertEqual(genomes, config["genomes"])
+
+            reads_1 = {
+                os.path.splitext(os.path.basename(r))[0].removesuffix(".1"): r
+                for r in SAMPLE_READS_FORWARD.split(" ")
+                }
+            self.assertEqual(reads_1, config["reads_1"])
+
+            reads_2 = {
+                os.path.splitext(os.path.basename(r))[0].removesuffix(".2"): r
+                for r in SAMPLE_READS_REVERSE.split(" ")
+                }
+            self.assertEqual(reads_2, config["reads_2"])
+
+            cluster_path = os.path.join("test", "coassemble", "target", "elusive_clusters.tsv")
+            self.assertTrue(os.path.exists(cluster_path))
+            expected = "\n".join(
+                [
+                    "\t".join([
+                        "samples",
+                        "length",
+                        "total_targets",
+                        "total_size",
+                        "recover_samples",
+                        "coassembly",
+                    ]),
+                    "\t".join([
+                        "sample_1,sample_3",
+                        "2",
+                        "1",
+                        "8456",
+                        "sample_1,sample_3",
+                        "coassembly_0"
+                    ]),
+                    ""
+                ]
+            )
+            with open(cluster_path) as f:
+                self.assertEqual(expected, f.read())
+
     def test_iterate_genome_input(self):
         with in_tempdir():
             cmd = (