Refactor to interface with MetabRef database

microbiomedata · Dec 13, 2024 · 6e884c3 · 6e884c3
1 parent f2a3ca8
commit 6e884c3
Show file tree

Hide file tree

Showing 6 changed files with 5 additions and 23 deletions.
diff --git a/configuration/gcms_corems.toml b/configuration/gcms_corems.toml
@@ -1,5 +1,5 @@
 [MolecularSearch]
-url_database = "sqlite:///db/MetabRef_Library_EILowRes_20240816.db"
+url_database = "sqlite:///db/MetabRef_Library_EILowRes_20240816.db"  # This is no longer needed
 ri_search_range = 35.0
 rt_search_range = 1.0
 correlation_threshold = 0.5

diff --git a/configuration/gcms_metams.toml b/configuration/gcms_metams.toml
@@ -1,5 +1,4 @@
 file_paths = [  "data/raw_data/GCMS_FAMES_01_GCMS-01_20191023.cdf", "data/raw_data/GCMS_FAMES_01_GCMS-01_20191023.cdf"]
-calibration_reference_path = "sqlite:///db/MetabRef_FAMEs_EILowRes_20240816.db"
 calibration_file_path = "data/raw_data/GCMS_FAMES_01_GCMS-01_20191023.cdf"
 corems_toml_path = "configuration/gcms_corems.toml"
 nmdc_metadata_path = "configuration/nmdc_metadata.json"

diff --git a/metaMS/cli.py b/metaMS/cli.py
@@ -25,7 +25,6 @@ def cli():
 
 @cli.command()
 @click.argument("file_paths", required=True, type=str)
-@click.argument("calibration_reference_path", required=True, type=str)
 @click.argument("calibration_file_path", required=True, type=str)
 @click.argument("output_directory", required=True, type=str)
 @click.argument("output_filename", required=True, type=str)
@@ -35,7 +34,6 @@ def cli():
 @click.option("--jobs", "-j", default=4, help="'cpu's'")
 def run_gcms_wdl_workflow(
     file_paths,
-    calibration_reference_path,
     calibration_file_path,
     output_directory,
     output_filename,
@@ -53,7 +51,6 @@ def run_gcms_wdl_workflow(
     click.echo("Running gcms workflow")
     run_gcms_metabolomics_workflow_wdl(
         file_paths,
-        calibration_reference_path,
         calibration_file_path,
         output_directory,
         output_filename,

diff --git a/metaMS/gcmsWorkflow.py b/metaMS/gcmsWorkflow.py
@@ -9,6 +9,7 @@
 from corems.mass_spectra.input.andiNetCDF import ReadAndiNetCDF
 from corems.molecular_id.factory.EI_SQL import EI_LowRes_SQLite
 from corems.molecular_id.search.compoundSearch import LowResMassSpectralMatch
+from corems.molecular_id.search.database_interfaces import MetabRefGCInterface
 
 
 @dataclass
@@ -81,8 +82,6 @@ def run_gcms_metabolomics_workflow_wdl(
     ----------
     file_paths : tuple(str)
         Paths to files to process.
-    calibration_reference_path : str
-        FAMEs retention index calibration reference filepath.
     calibration_file_path : str
         FAMEs retention index calibration filepath.
     output_directory : str
@@ -105,7 +104,6 @@ def run_gcms_metabolomics_workflow_wdl(
     # Store workflow parameters
     workflow_params = WorkflowParameters()
     workflow_params.file_paths = file_paths.split(",")
-    workflow_params.calibration_reference_path = calibration_reference_path
     workflow_params.calibration_file_path = calibration_file_path
     workflow_params.output_directory = output_directory
     workflow_params.output_filename = output_filename
@@ -130,9 +128,7 @@ def run_gcms_metabolomics_workflow_wdl(
     )
 
     # Load FAMEs calibration reference
-    fames_ref_sql = EI_LowRes_SQLite(
-        url=workflow_params.calibration_reference_path
-    )
+    fames_ref_sql = MetabRefGCInterface().get_fames(format="sql")
 
     # Compute RT:RI pairs
     rt_ri_pairs = get_rt_ri_pairs(gcms_cal_obj, sql_obj=fames_ref_sql)
@@ -257,9 +253,7 @@ def run_gcms_metabolomics_workflow(workflow_params_file, jobs):
     )
 
     # Load FAMEs calibration reference
-    fames_ref_sql = EI_LowRes_SQLite(
-        url=workflow_params.calibration_reference_path
-    )
+    fames_ref_sql = MetabRefGCInterface().get_fames(format="sql")
 
     # Compute RT:RI pairs
     rt_ri_pairs = get_rt_ri_pairs(gcms_cal_obj, sql_obj=fames_ref_sql)
@@ -360,19 +354,14 @@ def workflow_worker(args):
     # Unpack arguments
     file_path, ref_dict, corems_params_file, cal_file_path = args
 
-    # Load CoreMS parameters
-    corems_parameters = load_corems_parameters(corems_params_file)
-
     # Load data
     gcms = get_gcms(file_path, corems_params_file)
 
     # Calibrate retention indices
     gcms.calibrate_ri(ref_dict, cal_file_path)
 
     # Load reference database
-    ref_db_sql = EI_LowRes_SQLite(
-        url=corems_parameters["MolecularSearch"]["url_database"]
-    )
+    ref_db_sql = MetabRefGCInterface().get_library(format="sql")
 
     # Perform search
     lowResSearch = LowResMassSpectralMatch(gcms, sql_obj=ref_db_sql)

diff --git a/wdl/metaMS_gcms.wdl b/wdl/metaMS_gcms.wdl
@@ -13,7 +13,6 @@ workflow gcmsMetabolomics {
 task runMetaMSGCMS {
     input {
         Array[File] file_paths
-        File calibration_reference_path
         File calibration_file_path
         String output_directory
         String output_filename
@@ -26,7 +25,6 @@ task runMetaMSGCMS {
     command {
         metaMS run-gcms-wdl-workflow \
             ${sep=',' file_paths} \
-            ${calibration_reference_path} \
             ${calibration_file_path} \
             ${output_directory} \
             ${output_filename} \

diff --git a/wdl/metams_input_gcms.json b/wdl/metams_input_gcms.json
@@ -3,7 +3,6 @@
     "./data/raw_data/GCMS_FAMES_01_GCMS-01_20191023.cdf",
     "./data/raw_data/GCMS_FAMES_01_GCMS-01_20191023.cdf"
 ],
-"gcmsMetabolomics.runMetaMS.calibration_reference_path": "sqlite:///db/MetabRef_FAMEs_EILowRes_20240816.db",
 "gcmsMetabolomics.runMetaMS.calibration_file_path": "./data/raw_data/GCMS_FAMES_01_GCMS-01_20191023.cdf",
 "gcmsMetabolomics.runMetaMS.output_directory": "test_output",
 "gcmsMetabolomics.runMetaMS.output_filename": "test_dataset",