move params to the YAML files

CHANGELOG

@@ -1,9 +1,44 @@
 # Changelog
 
+## v1.2.0 rpetit3/sccmec "" 2024/08/25
+
+- Utilize both targets and full cassettes for classification
+- update default thresholds based on `camlhmp-blast-thresholds`
+    - `--min-targets-pident` 90
+    - `--min-targets-coverage` 80
+    - `--min-regions-pident` 85
+    - `--min-regions-coverage` 83
+- default values are set in YAMLs
+
 ## v1.1.0 rpetit3/sccmec "USA300" 2024/08/15
 
 - Update for latest camlhmp changes
 
 ## v1.0.0 rpetit3/sccmec "MRSA" 2024/04/30
 
 - Initial release
+
+@click.option(
+    "--min-targets-pident",
+    default=90,
+    show_default=True,
+    help="Minimum percent identity of targets to count a hit",
+)
+@click.option(
+    "--min-targets-coverage",
+    default=80,
+    show_default=True,
+    help="Minimum percent coverage of targets to count a hit",
+)
+@click.option(
+    "--min-regions-pident",
+    default=85,
+    show_default=True,
+    help="Minimum percent identity of regions to count a hit",
+)
+@click.option(
+    "--min-regions-coverage",
+    default=83,
+    show_default=True,
+    help="Minimum percent coverage of regions to count a hit",
+)

bin/sccmec

@@ -72,31 +72,31 @@ click.rich_click.OPTION_GROUPS = {
     "--yaml-targets",
     "-yt",
     required=True,
-    default=os.environ.get("CAML_YAML", None),
+    default=os.environ.get("SCCMEC_TARGETS_YAML", None),
     show_default=True,
     help="YAML file documenting the targets and types",
 )
 @click.option(
     "--yaml-regions",
     "-yr",
     required=True,
-    default=os.environ.get("CAML_YAML", None),
+    default=os.environ.get("SCCMEC_REGIONS_YAML", None),
     show_default=True,
     help="YAML file documenting the regions and types",
 )
 @click.option(
     "--targets",
     "-t",
     required=False if "--version" in sys.argv else True,
-    default=os.environ.get("CAML_TARGETS", None),
+    default=os.environ.get("SCCMEC_TARGETS_FASTA", None),
     show_default=True,
     help="Query targets in FASTA format",
 )
 @click.option(
     "--regions",
     "-r",
     required=False if "--version" in sys.argv else True,
-    default=os.environ.get("CAML_TARGETS", None),
+    default=os.environ.get("SCCMEC_REGIONS_FASTA", None),
     show_default=True,
     help="Query regions in FASTA format",
 )
@@ -118,25 +118,25 @@ click.rich_click.OPTION_GROUPS = {
 )
 @click.option(
     "--min-targets-pident",
-    default=95,
+    default=90,
     show_default=True,
     help="Minimum percent identity of targets to count a hit",
 )
 @click.option(
     "--min-targets-coverage",
-    default=95,
+    default=80,
     show_default=True,
     help="Minimum percent coverage of targets to count a hit",
 )
 @click.option(
     "--min-regions-pident",
-    default=95,
+    default=85,
     show_default=True,
     help="Minimum percent identity of regions to count a hit",
 )
 @click.option(
     "--min-regions-coverage",
-    default=95,
+    default=83,
     show_default=True,
     help="Minimum percent coverage of regions to count a hit",
 )
@@ -219,6 +219,21 @@ def sccmec(
     file_exists_error(regions_blast_tsv, force)
     file_exists_error(regions_details_tsv, force)
 
+    # Check if params are set in the YAML (only change if not set on the command line)
+    if "--min-targets-pident" not in sys.argv:
+        if "min_pident" in targets_framework["engine"]["params"]:
+            min_pident = targets_framework["engine"]["params"]["min_pident"]
+    if "--min-targets-coverage" not in sys.argv:
+        if "min_coverage" in targets_framework["engine"]["params"]:
+            min_coverage = targets_framework["engine"]["params"]["min_coverage"]
+
+    if "--min-regions-pident" not in sys.argv:
+        if "min_pident" in regions_framework["engine"]["params"]:
+            min_pident = regions_framework["engine"]["params"]["min_pident"]
+    if "--min-regions-coverage" not in sys.argv:
+        if "min_coverage" in regions_framework["engine"]["params"]:
+            min_coverage = regions_framework["engine"]["params"]["min_coverage"]
+
     # Describe the command line arguments
     console = rich.console.Console(stderr=True)
     print(

bin/sccmec-bioconda

@@ -0,0 +1,9 @@
+#!/usr/bin/env bash
+sccmec_dir=$(dirname $0)
+
+SCCMEC_TARGETS_YAML="${sccmec_dir}/../share/sccmec/sccmec-targets.yaml" \
+SCCMEC_TARGETS_FASTA="${sccmec_dir}/../share/sccmec/sccmec-targets.fasta" \
+SCCMEC_REGIONS_YAML="${sccmec_dir}/../share/sccmec/sccmec-regions.yaml" \
+SCCMEC_REGIONS_FASTA="${sccmec_dir}/../share/sccmec/sccmec-regions.fasta" \
+    sccmec-main \
+    "${@:1}"

data/sccmec-regions.yaml

@@ -13,6 +13,9 @@ metadata:
 engine:
   type: "blast" # The type of engine to use
   tool: blastn # The tool used to generate the data
+  params:
+    min_pident: 85    # The minimum percent identity for a hit
+    min_coverage: 83  # The minimum coverage for a hit
 
 # targets provides a list of sequence targets (primers, genes, proteins, etc...)
 targets:

data/sccmec-targets.yaml

@@ -13,6 +13,9 @@ metadata:
 engine:
   type: "blast" # The type of engine to use
   tool: blastn # The tool used to generate the data
+  params:
+    min_pident: 90    # The minimum percent identity for a hit
+    min_coverage: 80  # The minimum coverage for a hit
 
 # targets provides a list of sequence targets (primers, genes, proteins, etc...)
 targets: