Reorganize and rename config.yaml parameters

.test/targets.yaml

@@ -16,7 +16,7 @@ CONTEXT_FASTA:
   "data/context.fasta"
 OUTPUT_NAME:
   "test"
-DIVERSITY_BOOTSTRAP_REPS:
+DIVERSITY_REPS:
   10
 PLOTS:
   "../config/design_plots.R"

config/config.yaml

@@ -25,9 +25,12 @@ DEMIX:
 WINDOW:
   WIDTH: 1000
   STEP: 50
-GISAID_YAML:
-  "config/gisaid.yaml"
-DIVERSITY_BOOTSTRAP_REPS:
+GISAID:
+  CREDENTIALS: "config/gisaid.yaml"
+  DATE_COLUMN: "CollectionDate"
+  LOCATION_COLUMN: "ResidenceCity"
+  ACCESSION_COLUMN: "GISAIDEPI"
+DIVERSITY_REPS:
   1000
 LOG_PY_FMT:
   "%(asctime)s - %(name)s - %(levelname)-8s - %(message)s"

workflow/Snakefile

@@ -27,8 +27,8 @@ OUTDIR.mkdir(parents=True, exist_ok=True)
 ## Set context to be fetched automatically if CONTEXT_FASTA=null
 if "CONTEXT_FASTA" not in config.keys() or config["CONTEXT_FASTA"] is None:
     CONTEXT_FASTA = OUTDIR/"context"/"sequences.fasta"
-    if not Path(config["GISAID_YAML"]).is_file():
-        sys.exit(f"Tried to download a context dataset, but no GISAID credentials were found at '{config['GISAID_YAML']}' (see README.md).")
+    if not Path(config["GISAID"]["CREDENTIALS"]).is_file():
+        sys.exit(f"Tried to download a context dataset, but no GISAID credentials were found at '{config['GISAID']['CREDENTIALS']}' (see README.md).")
 elif Path(config["CONTEXT_FASTA"]).is_file():
     CONTEXT_FASTA = config["CONTEXT_FASTA"]
 else:

workflow/rules/context.smk

@@ -6,12 +6,12 @@ rule download_context:
         metadata = config["METADATA"],
         pango_report = OUTDIR/f"{OUTPUT_NAME}.lineage_report.csv"
     params:
-        gisaid_creds = config["GISAID_YAML"],
+        gisaid_creds = config["GISAID"]["CREDENTIALS"],
         date_window_span = 0.95,
         date_window_paddding_days = 14,
-        date_column = "CollectionDate",
-        location_column = "ResidenceCity",
-        samples_gisaid_accession_column = "GISAIDEPI",
+        date_column = config["GISAID"]["DATE_COLUMN"],
+        location_column = config["GISAID"]["LOCATION_COLUMN"],
+        samples_gisaid_accession_column = config["GISAID"]["ACCESSION_COLUMN"],
         context_gisaid_accession_column = "accession_id",
         host = "Human",
         exclude_low_coverage = True,
@@ -21,7 +21,7 @@ rule download_context:
         min_sleep = 1,
         max_sleep = 3,
         chunk_length = 3000,
-        min_theoretical_combinations = config["DIVERSITY_BOOTSTRAP_REPS"]
+        min_theoretical_combinations = config["DIVERSITY_REPS"]
     output:
         fasta = temp(OUTDIR/"context"/"sequences.fasta"),
         metadata = temp(OUTDIR/"context"/"metadata.csv"),

workflow/rules/report.smk

@@ -33,7 +33,7 @@ rule diversity:
     conda: "../envs/renv.yaml"
     params:
         design = config["PLOTS"],
-        bootstrap_reps = config["DIVERSITY_BOOTSTRAP_REPS"],
+        bootstrap_reps = config["DIVERSITY_REPS"],
         plot_width = 159.2,
         plot_height = 119.4
     input: