allow params in the yaml

CHANGELOG.md

@@ -1,6 +1,16 @@
 
 # Changelog
 
+## v1.0.2 rpetit3/camlhmp "Wild Bactrian" 2024/08/25
+
+### `Added`
+
+- allow `min-coverage` and `min-coverage` params to be set in the yaml
+    - Priority:
+        - user provided command line arguments (`--min-pident` and `--min-coverage`)
+        - schema provided values
+        - default values specified in the command script
+
 ## v1.0.1 rpetit3/camlhmp "Bactrian" 2024/08/25
 
 ### `Added`

camlhmp/cli/blast/alleles.py

@@ -168,10 +168,19 @@ def camlhmp_blast_alleles(
     file_exists_error(blast_tsv, force)
     file_exists_error(details_tsv, force)
 
+    # Check if params are set in the YAML (only change if not set on the command line)
+    if "--min-pident" not in sys.argv:
+        if "min_pident" in framework["engine"]["params"]:
+            min_pident = framework["engine"]["params"]["min_pident"]
+    if "--min-coverage" not in sys.argv:
+        if "min_coverage" in framework["engine"]["params"]:
+            min_coverage = framework["engine"]["params"]["min_coverage"]
+
+
     # Describe the command line arguments
     console = rich.console.Console(stderr=True)
     print(
-        "[italic]Running [deep_sky_blue1]camlhmp[/deep_sky_blue1] with following parameters:[/italic]",
+        "[italic]Running [deep_sky_blue1]camlhmp-blast-alleless[/deep_sky_blue1] with following parameters:[/italic]",
         file=sys.stderr,
     )
     print(f"[italic]    --input {input}[/italic]", file=sys.stderr)

camlhmp/cli/blast/regions.py

@@ -168,6 +168,14 @@ def camlhmp_blast_regions(
     file_exists_error(blast_tsv, force)
     file_exists_error(details_tsv, force)
 
+    # Check if params are set in the YAML (only change if not set on the command line)
+    if "--min-pident" not in sys.argv:
+        if "min_pident" in framework["engine"]["params"]:
+            min_pident = framework["engine"]["params"]["min_pident"]
+    if "--min-coverage" not in sys.argv:
+        if "min_coverage" in framework["engine"]["params"]:
+            min_coverage = framework["engine"]["params"]["min_coverage"]
+
     # Describe the command line arguments
     console = rich.console.Console(stderr=True)
     print(

camlhmp/cli/blast/targets.py

@@ -168,6 +168,15 @@ def camlhmp_blast_targets(
     file_exists_error(blast_tsv, force)
     file_exists_error(details_tsv, force)
 
+    # Check if params are set in the YAML (only change if not set on the command line)
+    if "--min-pident" not in sys.argv:
+        if "min_pident" in framework["engine"]["params"]:
+            min_pident = framework["engine"]["params"]["min_pident"]
+    if "--min-coverage" not in sys.argv:
+        if "min_coverage" in framework["engine"]["params"]:
+            min_coverage = framework["engine"]["params"]["min_coverage"]
+
+
     # Describe the command line arguments
     console = rich.console.Console(stderr=True)
     print(

docs/schema.md

@@ -21,8 +21,11 @@ metadata:
 # engine: specifies the computational tools and additional parameters used for sequence
 #         analysis.
 engine:
-  type: ""        # The type of tool used to generate the data
-  tool: ""        # The tool used to generate the data
+  type: ""            # The type of tool used to generate the data
+  tool: ""            # The tool used to generate the data
+  params: {}          # Additional parameters for the tool
+    min_pident: int   # Minimum percent identity for the tool
+    min_coverage: int # Minimum percent coverage for the tool
 
 # targets: Lists the specific sequence targets such as genes, proteins, or markers that the
 #          schema will analyze. These should be included in the associated sequence query data
@@ -66,10 +69,11 @@ The `metadata` section provides general information about the schema. This inclu
 
 The `engine` section specifies the computational tools used for sequence analysis.
 
-| Field | Type   | Description                                      |
-|-------|--------|--------------------------------------------------|
-| type  | string | The type of engine used for analysis             |
-| tool  | string | The specific tool to be used for the engine      |
+| Field  | Type   | Description                                          |
+|--------|--------|------------------------------------------------------|
+| type   | string | The type of engine used for analysis                 |
+| tool   | string | The specific tool to be used for the engine          |
+| params | dict   | Additional parameters for the tool to use as default |
 
 ## targets
 

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "camlhmp"
-version = "1.0.1"
+version = "1.0.2"
 description = "Classification through yAML Heuristic Mapping Protocol"
 authors = [
     "Robert A. Petit III <[email protected]>",