Merge branch 'di_workflow_bjm' into 'LC_FTICR_Final'

Direct Infusion workflow updates and improvements

See merge request mass-spectrometry/enviroms!10

Author: corilo

.gitignore

@@ -39,4 +39,9 @@ share/python-wheels/
 *.egg
 
 # Raw data files
-*.raw
+*.raw
+*.d
+data/raw_data/example_data/
+
+# Processed data
+output

Makefile

@@ -72,8 +72,7 @@ docker-build:
 
 docker-build-local:
 
-	docker build -t local-enviroms:$(version) .
-
+	docker build --pull --no-cache -t local-enviroms:latest .
 
 docker-run-di:
 
@@ -91,14 +90,25 @@ cascade-run:
 
 	srun -A mscms -t 240 -N 1 -n time enviroMS run-di -r 2 --mpi  /dtemp/mscms/enviroms/data/configuration/enviroms.toml
 
+cli-run-di :
+	enviroMS run_di configuration/di_enviroms.toml --jobs 1 --replicas 1 --tasks 1
+
 wdl-run-di :
 
 	miniwdl run wdl/di_fticr_ms.wdl -i wdl/di_fticr_wdl_input.json --verbose --no-cache --copy-input-files
 
+wdl-run-di-local :
+	@make docker-build-local
+	miniwdl run wdl/di_fticr_ms.wdl -i wdl/di_fticr_wdl_input_local_docker.json --verbose --no-cache --copy-input-files
+
 wdl-run-lc :
 
 	miniwdl run wdl/lc_fticr_ms.wdl -i wdl/lc_fticr_wdl_input.json --verbose --no-cache --copy-input-files
 
+wdl-run-lc-local:
+
+	miniwdl run wdl/lc_fticr_ms.wdl -i wdl/lc_fticr_wdl_input_local_docker.json --verbose --no-cache --copy-input-files
+
 get-lcms-fticr-test-data:
 
 	@echo "Downloading test files for LC-MS FT-ICR workflow"
@@ -132,6 +142,4 @@ get-lcms-fticr-test-data:
 	else echo "Reference file exists"; fi
 	@echo "LC-MS FT-ICR test files complete"
 
-wdl-run-lc-local:
 
-	miniwdl run wdl/lc_fticr_ms.wdl -i wdl/lc_fticr_wdl_input_local_docker.json --verbose --no-cache --copy-input-files

README.md

@@ -199,7 +199,7 @@ A docker image containing the EnviroMS command line as code entry-point
 - Run Workflow from Container:
 
     $(data_dir) = dir_containing the FT-ICR MS data
-    $(configuration_dir) = dir_containing the enviroms.toml, corems.toml and nmdc_metadata.json
+    $(configuration_dir) = dir_containing the enviroms.toml and corems.toml
     
     ```bash
     docker run -v $(data_dir):/enviroms/data \

configuration/di_corems.toml

@@ -5,7 +5,7 @@ isotopologue_filter_atoms = [ "Cl", "Br",]
 use_runtime_kendrick_filter = false
 use_min_peaks_filter = true
 min_peaks_per_class = 15
-url_database = "sqlite:///db/molformulas.db"
+url_database = ""
 db_jobs = 3
 db_chunk_size = 300
 ion_charge = -1
@@ -77,14 +77,17 @@ implemented_kendrick_rounding_methods = [ "floor", "ceil", "round",]
 peak_derivative_threshold = 0.0
 peak_min_prominence_percent = 0.1
 min_peak_datapoints = 5.0
-peak_max_prominence_percent = 0.1
+peak_max_prominence_percent = 1
 peak_height_max_percent = 10.0
 legacy_resolving_power = true
 centroid_legacy_polyfit = false
 
 [MolecularFormulaSearch.usedAtoms]
 C = [ 1, 100,]
-H = [ 1, 200,]
+H = [ 4, 200,]
+O = [ 0, 22,]
+N = [ 0, 1,]
+S = [ 0, 1,]
 
 [MolecularFormulaSearch.used_atom_valences]
 C = 4

configuration/di_enviroms.toml

@@ -1,18 +1,19 @@
 raw_file_start_scan = 1
 raw_file_final_scan = 7
-file_paths = [ "data/raw_data/20190709_WK_CADY_Auto_S16_H1_Post_O5_1_01_36.tsv" ]
+file_paths = ["data/raw_data/NEG_ESI_SRFA_Auto.d", "data/raw_data/SRFA_40ppm_NegativeESI_21T.raw"]
 output_directory = "output"
 output_group_name = "..."
 output_type = "csv"
 polarity = -1
 is_centroid = true
-corems_toml_path = "configuration/corems.toml"
-nmdc_metadata_path = "configuration/nmdc_metadata.json"
+corems_toml_path = "configuration/di_corems.toml"
 calibrate = true
-calibration_ref_file_path = "data/raw_data/SRFA.ref"
+batch_calibrate = true
+calibration_ref_file_path = "./data/reference/Hawkes_neg.ref"
 plot_mz_error = true
 plot_ms_assigned_unassigned = true
 plot_c_dbe = true
 plot_van_krevelen = true
 plot_ms_classes = true
-plot_mz_error_classes = true
+plot_mz_error_classes = true
+plot_qc = true