Merge pull request #1452 from Clinical-Genomics/release_v15.0.1

.gitignore

@@ -35,3 +35,4 @@ BALSAMIC_reference/
 BALSAMIC/workflows/umi_workflow
 BALSAMIC/workflows/umi_workflow/.*
 .idea/
+.vscode/

BALSAMIC/commands/config/case.py

@@ -1,4 +1,5 @@
 """Balsamic config case CLI."""
+
 import logging
 from datetime import datetime
 from pathlib import Path
@@ -36,6 +37,7 @@
     OPTION_SWEGEN_SV,
     OPTION_TUMOR_SAMPLE_NAME,
     OPTION_UMI,
+    OPTION_UMI_MIN_READS,
     OPTION_UMI_TRIM_LENGTH,
 )
 from BALSAMIC.constants.analysis import BIOINFO_TOOL_ENV, AnalysisWorkflow, Gender
@@ -87,6 +89,7 @@
 @OPTION_TUMOR_SAMPLE_NAME
 @OPTION_UMI
 @OPTION_UMI_TRIM_LENGTH
+@OPTION_UMI_MIN_READS
 @click.pass_context
 def case_config(
     context: click.Context,
@@ -119,6 +122,7 @@ def case_config(
     tumor_sample_name: str,
     umi: bool,
     umi_trim_length: int,
+    umi_min_reads: str | None,
 ):
     references_path: Path = Path(balsamic_cache, cache_version, genome_version)
     references: Dict[str, Path] = get_absolute_paths_dict(
@@ -205,6 +209,7 @@ def case_config(
             "analysis_workflow": analysis_workflow,
             "config_creation_date": datetime.now().strftime("%Y-%m-%d %H:%M"),
         },
+        custom_filters={"umi_min_reads": umi_min_reads if umi_min_reads else None},
         reference=references,
         singularity={
             "image": Path(balsamic_cache, cache_version, "containers").as_posix()
@@ -221,14 +226,16 @@ def case_config(
             bioinfo_tools=BIOINFO_TOOL_ENV,
             container_conda_env_path=CONTAINERS_DIR,
         ),
-        panel={
-            "exome": exome,
-            "capture_kit": panel_bed,
-            "chrom": get_panel_chrom(panel_bed),
-            "pon_cnn": pon_cnn,
-        }
-        if panel_bed
-        else None,
+        panel=(
+            {
+                "exome": exome,
+                "capture_kit": panel_bed,
+                "chrom": get_panel_chrom(panel_bed),
+                "pon_cnn": pon_cnn,
+            }
+            if panel_bed
+            else None
+        ),
     ).model_dump(by_alias=True, exclude_none=True)
     LOG.info("Balsamic config model instantiated successfully")
 

BALSAMIC/commands/options.py

@@ -1,4 +1,5 @@
 """Balsamic command options."""
+
 import click
 
 from BALSAMIC import __version__ as balsamic_version
@@ -22,7 +23,11 @@
 from BALSAMIC.constants.constants import LOG_LEVELS, LogLevel
 from BALSAMIC.constants.rules import DELIVERY_RULES
 from BALSAMIC.constants.workflow_params import VCF_DICT
-from BALSAMIC.utils.cli import validate_cache_version, validate_exome_option
+from BALSAMIC.utils.cli import (
+    validate_cache_version,
+    validate_exome_option,
+    validate_umi_min_reads,
+)
 
 OPTION_ADAPTER_TRIM = click.option(
     "--adapter-trim/--no-adapter-trim",
@@ -423,3 +428,10 @@
     type=click.INT,
     help="Trim N bases from reads in FASTQ file",
 )
+
+OPTION_UMI_MIN_READS = click.option(
+    "--umi-min-reads",
+    type=click.STRING,
+    callback=validate_umi_min_reads,
+    help="Minimum raw reads supporting each UMI group. Format: 'x,y,z'.",
+)

BALSAMIC/models/config.py

@@ -1,4 +1,5 @@
 """Balsamic analysis config case models."""
+
 import re
 from glob import glob
 from pathlib import Path
@@ -171,6 +172,12 @@ def validate_pon_version(cls, pon_version: Optional[str]):
         return pon_version
 
 
+class CustomFilters(BaseModel):
+    """Variant calling custom filters."""
+
+    umi_min_reads: str | None = None
+
+
 class ConfigModel(BaseModel):
     """
     Class providing common functions and variables for different balsamic workflows.
@@ -186,6 +193,7 @@ class ConfigModel(BaseModel):
         vcf : Field(VCFmodel); variables relevant for variant calling pipeline
         background_variants: Field(Path(optional)); path to BACKGROUND VARIANTS for UMI
         analysis: Field(AnalysisModel); Pydantic model containing workflow variables
+        custom_filters: Field(CustomFilters); custom parameters for variant filtering
 
     This class also contains functions that help retrieve sample and file information,
     facilitating BALSAMIC run operations in Snakemake.
@@ -211,6 +219,7 @@ class ConfigModel(BaseModel):
     vcf: Optional[VCFModel] = None
     background_variants: Optional[str] = None
     analysis: AnalysisModel
+    custom_filters: CustomFilters | None = None
 
     @field_validator("reference")
     def abspath_as_str(cls, reference: Dict[str, Path]):

BALSAMIC/utils/cli.py

@@ -296,12 +296,16 @@ def get_fastq_info(sample_name: str, fastq_path: str) -> Dict[str, FastqInfoMode
                 "fwd": fwd_fastq,
                 "rev": rev_fastq,
             }
-            fastq_dict[fastq_pair_pattern].update(
-                {
-                    "fwd_resolved": Path(fwd_fastq).resolve().as_posix(),
-                    "rev_resolved": Path(rev_fastq).resolve().as_posix(),
-                }
-            ) if Path(fwd_fastq).is_symlink() or Path(rev_fastq).is_symlink() else None
+            (
+                fastq_dict[fastq_pair_pattern].update(
+                    {
+                        "fwd_resolved": Path(fwd_fastq).resolve().as_posix(),
+                        "rev_resolved": Path(rev_fastq).resolve().as_posix(),
+                    }
+                )
+                if Path(fwd_fastq).is_symlink() or Path(rev_fastq).is_symlink()
+                else None
+            )
 
     if not fastq_dict:
         error_message = f"No fastqs found for: {sample_name} in {fastq_path}"
@@ -511,3 +515,16 @@ def validate_cache_version(
     raise click.BadParameter(
         f"Invalid cache version format. Use '{CacheVersion.DEVELOP}' or 'X.X.X'."
     )
+
+
+def validate_umi_min_reads(
+    _ctx: click.Context, _param: click.Parameter, umi_min_reads: str | None
+) -> str:
+    """Validate the provided cache version."""
+    if umi_min_reads:
+        umi_min_reads_parts: List[str] = umi_min_reads.split(",")
+        if len(umi_min_reads_parts) == 3 and all(
+            part.isdigit() for part in umi_min_reads_parts
+        ):
+            return umi_min_reads
+        raise click.BadParameter("Invalid UMI minimum reads format. Use 'x,y,z'.")