Add GBM and remove GMM.

meteor/phylogeny.py

@@ -123,10 +123,11 @@ def execute(self) -> None:
         if parse(raxml_ng_version) < self.meteor.MIN_RAXML_NG_VERSION:
             logging.error(
                 "The raxml-ng version %s is outdated for meteor. Please update raxml-ng to >= %s.",
-                raxml_ng_version, self.meteor.MIN_RAXML_NG_VERSION
+                raxml_ng_version,
+                self.meteor.MIN_RAXML_NG_VERSION,
             )
             sys.exit(1)
-            
+
         # Start phylogenies
         start = perf_counter()
         tree_files: list[Path] = []
@@ -156,7 +157,7 @@ def execute(self) -> None:
                             "--model",
                             "GTR+G",
                             "--redo",
-                            # "--force perf_threads", # only with raxml-ng-mpi
+                            "--force perf_threads",  # not working with raxml-ng-mpi
                             "--prefix",
                             str(tree_file.resolve()),
                         ]

meteor/profiler.py

@@ -158,7 +158,7 @@ def __post_init__(self):
 
             # Initialize the module definition file
             self.module_path = (
-                importlib.resources.files("meteor") / "data/GMM_definition.tsv"
+                importlib.resources.files("meteor") / "data/modules_definition.tsv"
             )
             assert self.module_path.is_file()
 

meteor/tests/test_merging.py

@@ -118,7 +118,7 @@ def test_extract_json_info(merging_profiles: Merging) -> None:
     )
     assert info == {
         "msp_filter": 0.1,
-        "modules_def": "GMM_definition.tsv",
+        "modules_def": "modules_definition.tsv",
         "bowtie_file": "sample1.sam",
     }
 

meteor/tests/test_merging/expected_output/test_project_census_stage_2_report.tsv

@@ -1,4 +1,4 @@
 sample	sample_name	census_status	full_sample_name	fastq_file	mapping_tool	mapping_tool_version	mapping_date	reference_name	trim	alignment_number	mapping_type	meteor_version	identity_threshold	total_read_count	mapped_read_count	overall_alignment_rate	gene_count	msp_count	msp_signal	mustard_signal	fastq_files	database_type	bowtie_file	profiling_date	normalization	rarefaction_level	seed	msp_core_size	msp_filter	msp_def	mustard_filename	modules_db	modules_db_filenames	modules_def	module_completeness
-sample1	sample1	0	sample1_trimmed.Q17.converted.noHost	sample1_trimmed.Q17.converted.noHost.fastq.gz	bowtie2	2.5.1	2023-11-17	IGC2	80	10000	end-to-end	2.0.9	0.95	19234567	16234987	84.41	627516	297	0.63	0.0	['fastq1.fastq.gz', 'fastq2.fastq.gz']	complete	sample1.sam	2023-11-17	fpkm	5000000	1234	100	0.1	IGC2_1990MSPs.tsv	IGC2_mustard.tsv	kegg	IGC2_kegg_107.tsv	GMM_definition.tsv	0.9
-sample2	sample2	0	sample2_trimmed.Q17.converted.noHost	sample2_trimmed.Q17.converted.noHost.fastq.gz	bowtie2	2.5.1	2023-11-17	IGC2	80	10000	end-to-end	2.0.9	0.95	15000000	10000000	66.67	687432	325	0.87	0.1	['fastq1.fastq.gz', 'fastq2.fastq.gz']	complete	sample2.sam	2023-11-17	fpkm	5000000	1234	100	0.1	IGC2_1990MSPs.tsv	IGC2_mustard.tsv	kegg	IGC2_kegg_107.tsv	GMM_definition.tsv	0.9
-sample3	sample3	0	sample3_trimmed.Q17.converted.noHost	sample3_trimmed.Q17.converted.noHost.fastq.gz	bowtie2	2.5.1	2023-11-17	IGC2	80	10000	end-to-end	2.0.9	0.95	20000000	10000000	50.00	599999	354	0.56	0.3	['fastq1.fastq.gz', 'fastq2.fastq.gz']	complete	sample3.sam	2023-11-17	fpkm	5000000	1234	90	0.1	IGC2_1990MSPs.tsv	IGC2_mustard.tsv	kegg	IGC2_kegg_107.tsv	GMM_definition.tsv	0.9
+sample1	sample1	0	sample1_trimmed.Q17.converted.noHost	sample1_trimmed.Q17.converted.noHost.fastq.gz	bowtie2	2.5.1	2023-11-17	IGC2	80	10000	end-to-end	2.0.9	0.95	19234567	16234987	84.41	627516	297	0.63	0.0	['fastq1.fastq.gz', 'fastq2.fastq.gz']	complete	sample1.sam	2023-11-17	fpkm	5000000	1234	100	0.1	IGC2_1990MSPs.tsv	IGC2_mustard.tsv	kegg	IGC2_kegg_107.tsv	modules_definition.tsv	0.9
+sample2	sample2	0	sample2_trimmed.Q17.converted.noHost	sample2_trimmed.Q17.converted.noHost.fastq.gz	bowtie2	2.5.1	2023-11-17	IGC2	80	10000	end-to-end	2.0.9	0.95	15000000	10000000	66.67	687432	325	0.87	0.1	['fastq1.fastq.gz', 'fastq2.fastq.gz']	complete	sample2.sam	2023-11-17	fpkm	5000000	1234	100	0.1	IGC2_1990MSPs.tsv	IGC2_mustard.tsv	kegg	IGC2_kegg_107.tsv	modules_definition.tsv	0.9
+sample3	sample3	0	sample3_trimmed.Q17.converted.noHost	sample3_trimmed.Q17.converted.noHost.fastq.gz	bowtie2	2.5.1	2023-11-17	IGC2	80	10000	end-to-end	2.0.9	0.95	20000000	10000000	50.00	599999	354	0.56	0.3	['fastq1.fastq.gz', 'fastq2.fastq.gz']	complete	sample3.sam	2023-11-17	fpkm	5000000	1234	90	0.1	IGC2_1990MSPs.tsv	IGC2_mustard.tsv	kegg	IGC2_kegg_107.tsv	modules_definition.tsv	0.9

meteor/tests/test_merging/profiles/sample1/sample1_census_stage_2.json

@@ -42,7 +42,7 @@
     "mustard_filename": "IGC2_mustard.tsv",
     "modules_db": "kegg",
     "modules_db_filenames": "IGC2_kegg_107.tsv",
-    "modules_def": "GMM_definition.tsv",
+    "modules_def": "modules_definition.tsv",
     "module_completeness": 0.9
   },
   "profiling_stats": {

meteor/tests/test_merging/profiles/sample2/sample2_census_stage_2.json

@@ -42,7 +42,7 @@
     "mustard_filename": "IGC2_mustard.tsv",
     "modules_db": "kegg",
     "modules_db_filenames": "IGC2_kegg_107.tsv",
-    "modules_def": "GMM_definition.tsv",
+    "modules_def": "modules_definition.tsv",
     "module_completeness": 0.9
   },
   "profiling_stats": {

meteor/tests/test_merging/profiles/sample3/sample3_census_stage_2.json

@@ -42,7 +42,7 @@
     "mustard_filename": "IGC2_mustard.tsv",
     "modules_db": "kegg",
     "modules_db_filenames": "IGC2_kegg_107.tsv",
-    "modules_def": "GMM_definition.tsv",
+    "modules_def": "modules_definition.tsv",
     "module_completeness": 0.9
   },
   "profiling_stats": {