merge develop

BALSAMIC/commands/config/case.py

@@ -31,6 +31,8 @@
     OPTION_NORMAL_SAMPLE_NAME,
     OPTION_PANEL_BED,
     OPTION_PON_CNN,
+    OPTION_SENTIEON_INSTALL_DIR,
+    OPTION_SENTIEON_LICENSE,
     OPTION_SWEGEN_SNV,
     OPTION_SWEGEN_SV,
     OPTION_TUMOR_SAMPLE_NAME,
@@ -39,7 +41,11 @@
 from BALSAMIC.constants.analysis import BIOINFO_TOOL_ENV, AnalysisWorkflow, Gender
 from BALSAMIC.constants.cache import GenomeVersion
 from BALSAMIC.constants.constants import FileType
-from BALSAMIC.constants.paths import CONTAINERS_DIR
+from BALSAMIC.constants.paths import (
+    CONTAINERS_DIR,
+    SENTIEON_DNASCOPE_MODEL,
+    SENTIEON_TNSCOPE_MODEL,
+)
 from BALSAMIC.constants.workflow_params import VCF_DICT
 from BALSAMIC.models.config import ConfigModel
 from BALSAMIC.utils.cli import (
@@ -78,6 +84,8 @@
 @OPTION_NORMAL_SAMPLE_NAME
 @OPTION_PANEL_BED
 @OPTION_PON_CNN
+@OPTION_SENTIEON_INSTALL_DIR
+@OPTION_SENTIEON_LICENSE
 @OPTION_SWEGEN_SNV
 @OPTION_SWEGEN_SV
 @OPTION_TUMOR_SAMPLE_NAME
@@ -107,6 +115,8 @@ def case_config(
     normal_sample_name: str,
     panel_bed: Path,
     pon_cnn: Path,
+    sentieon_install_dir: Path,
+    sentieon_license: str,
     swegen_snv: Path,
     swegen_sv: Path,
     tumor_sample_name: str,
@@ -177,6 +187,13 @@ def case_config(
 
     config_collection_dict = ConfigModel(
         QC={},
+        sentieon={
+            "sentieon_install_dir": sentieon_install_dir,
+            "sentieon_license": sentieon_license,
+            "sentieon_exec": Path(sentieon_install_dir, "bin", "sentieon").as_posix(),
+            "dnascope_model": SENTIEON_DNASCOPE_MODEL.as_posix(),
+            "tnscope_model": SENTIEON_TNSCOPE_MODEL.as_posix(),
+        },
         analysis={
             "case_id": case_id,
             "gender": gender,

BALSAMIC/commands/config/pon.py

@@ -15,14 +15,20 @@
     OPTION_FASTQ_PATH,
     OPTION_GENOME_INTERVAL,
     OPTION_GENOME_VERSION,
+    OPTION_SENTIEON_INSTALL_DIR,
+    OPTION_SENTIEON_LICENSE,
     OPTION_PANEL_BED,
     OPTION_PON_VERSION,
     OPTION_PON_WORKFLOW,
 )
 from BALSAMIC.constants.analysis import BIOINFO_TOOL_ENV, PONWorkflow
 from BALSAMIC.constants.cache import GenomeVersion
 from BALSAMIC.constants.constants import FileType
-from BALSAMIC.constants.paths import CONTAINERS_DIR
+from BALSAMIC.constants.paths import (
+    CONTAINERS_DIR,
+    SENTIEON_DNASCOPE_MODEL,
+    SENTIEON_TNSCOPE_MODEL,
+)
 from BALSAMIC.models.config import ConfigModel
 from BALSAMIC.utils.cli import (
     generate_graph,
@@ -44,6 +50,8 @@
 @OPTION_FASTQ_PATH
 @OPTION_GENOME_VERSION
 @OPTION_GENOME_INTERVAL
+@OPTION_SENTIEON_INSTALL_DIR
+@OPTION_SENTIEON_LICENSE
 @OPTION_PANEL_BED
 @OPTION_PON_WORKFLOW
 @OPTION_PON_VERSION
@@ -57,6 +65,8 @@ def pon_config(
     fastq_path: Path,
     genome_version: GenomeVersion,
     genome_interval: Path,
+    sentieon_install_dir: Path,
+    sentieon_license: str,
     panel_bed: Path,
     pon_workflow: PONWorkflow,
     version: str,
@@ -94,6 +104,13 @@ def pon_config(
 
     config_collection_dict = ConfigModel(
         QC={},
+        sentieon={
+            "sentieon_install_dir": sentieon_install_dir,
+            "sentieon_license": sentieon_license,
+            "sentieon_exec": Path(sentieon_install_dir, "bin", "sentieon").as_posix(),
+            "dnascope_model": SENTIEON_DNASCOPE_MODEL.as_posix(),
+            "tnscope_model": SENTIEON_TNSCOPE_MODEL.as_posix(),
+        },
         analysis={
             "case_id": case_id,
             "analysis_dir": analysis_dir,

BALSAMIC/commands/options.py

@@ -354,6 +354,20 @@
     help="Sample configuration file",
 )
 
+OPTION_SENTIEON_INSTALL_DIR = click.option(
+    "--sentieon-install-dir",
+    type=click.Path(exists=True, resolve_path=True),
+    required=True,
+    help="Path to Sentieon install directory",
+)
+
+OPTION_SENTIEON_LICENSE = click.option(
+    "--sentieon-license",
+    required=True,
+    type=click.STRING,
+    help="Sentieon license in format IP:Port",
+)
+
 OPTION_SHOW_ONLY_MISSING_FILES = click.option(
     "-m",
     "--show-only-missing",

BALSAMIC/constants/paths.py

@@ -13,10 +13,10 @@
 
 # Sentieon specific constants
 SENTIEON_MODELS_DIR: Path = Path(BALSAMIC_DIR, "assets", "sentieon_models")
-SENTIEON_DNASCOPE_DIR: Path = Path(
+SENTIEON_DNASCOPE_MODEL: Path = Path(
     SENTIEON_MODELS_DIR, "SentieonDNAscopeModelBeta0.4a-201808.05.model"
 )
-SENTIEON_TNSCOPE_DIR: Path = Path(
+SENTIEON_TNSCOPE_MODEL: Path = Path(
     SENTIEON_MODELS_DIR, "SentieonTNscopeModel_GiAB_HighAF_LowFP-201711.05.model"
 )
 

BALSAMIC/containers/cnvkit/Dockerfile

@@ -1,34 +1,32 @@
 FROM python:3.10-slim
 
-LABEL base.image="python:3.10-slim"
-LABEL maintainer="Clinical Genomics"
-LABEL about.contact="support@clinicalgenomics.se"
-LABEL software="cnvkit"
-LABEL software.version="0.9.10"
-LABEL about.summary="Copy number variant detection from targeted DNA sequencing"
-LABEL about.home="https://github.com/etal/cnvkit"
-LABEL about.documentation="https://cnvkit.readthedocs.io"
-LABEL about.license="MIT License (MIT)"
+LABEL base.image="python:3.10-slim" \
+    maintainer="Clinical Genomics" \
+    about.contact="support@clinicalgenomics.se" \
+    software="CNVkit" \
+    software.version="0.9.10" \
+    about.summary="Copy number variant detection from targeted DNA sequencing" \
+    about.home="https://github.com/etal/cnvkit" \
+    about.documentation="https://cnvkit.readthedocs.io" \
+    about.license="MIT License (MIT)"
 
 ENV DEBIAN_FRONTEND noninteractive
+ENV VENV /opt/venv
+ENV PATH="${VENV}/bin:$PATH"
+
 RUN apt-get update && apt-get -y upgrade && \
     apt-get -y install --no-install-recommends tabix liblzma-dev zlib1g-dev \
     r-base-core r-bioc-dnacopy && \
     apt-get clean && \
     rm -rf /var/lib/apt/lists/*
 
-ENV VENV /opt/venv
-
-RUN python -m pip install --upgrade --no-cache-dir pip
-RUN python -m venv ${VENV}
-ENV PATH="${VENV}/bin:$PATH"
-
-RUN pip install --no-cache-dir --upgrade pip
-
-RUN pip install --no-cache-dir cnvkit==0.9.10
+RUN python -m pip install --upgrade --no-cache-dir pip && \
+    python -m venv ${VENV} && \
+    pip install --no-cache-dir "cnvkit==0.9.10" "numpy<2.0.0"
 
-RUN adduser --disabled-password --gecos '' ubuntu && \
-    chsh -s /bin/bash && mkdir -p /home/ubuntu
+RUN adduser --disabled-password --gecos "" ubuntu && \
+    chsh -s /bin/bash ubuntu && \
+    mkdir -p /home/ubuntu
 
 USER ubuntu
 WORKDIR /home/ubuntu

BALSAMIC/models/config.py

@@ -178,6 +178,23 @@ class CustomFilters(BaseModel):
     umi_min_reads: str | None = None
 
 
+class Sentieon(BaseModel):
+    """
+    Class providing common functions and variables for different balsamic workflows.
+
+    Attributes:
+        sentieon_install_dir: Field(required); path to Sentieon installation directory
+        sentieon_exec:  Field(required); path to Sentieon executeable
+        sentieon_license: Field(required); Sentieon license string
+    """
+
+    sentieon_install_dir: Annotated[str, AfterValidator(is_dir)]
+    sentieon_exec: Annotated[str, AfterValidator(is_file)]
+    sentieon_license: str
+    dnascope_model: Annotated[str, AfterValidator(is_file)]
+    tnscope_model: Annotated[str, AfterValidator(is_file)]
+
+
 class ConfigModel(BaseModel):
     """
     Class providing common functions and variables for different balsamic workflows.
@@ -194,6 +211,7 @@ class ConfigModel(BaseModel):
         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
+        sentieon: Field(required); Sentieon model attributes
 
     This class also contains functions that help retrieve sample and file information,
     facilitating BALSAMIC run operations in Snakemake.
@@ -220,6 +238,7 @@ class ConfigModel(BaseModel):
     background_variants: Optional[str] = None
     analysis: AnalysisModel
     custom_filters: CustomFilters | None = None
+    sentieon: Sentieon
 
     @field_validator("reference")
     def abspath_as_str(cls, reference: Dict[str, Path]):

BALSAMIC/snakemake_rules/align/tga_bam_postprocess.rule

@@ -14,8 +14,8 @@ rule sentieon_dedup_consensus:
         Path(benchmark_dir, "sentieon_dedup_{sample_type}.{sample}.tsv").as_posix()
     params:
         tmpdir = tempfile.mkdtemp(prefix=tmp_dir),
-        sentieon_exec = config["SENTIEON_EXEC"],
-        sentieon_lic = config["SENTIEON_LICENSE"],
+        sentieon_exec = config_model.sentieon.sentieon_exec,
+        sentieon_lic = config_model.sentieon.sentieon_license,
         sample_id = "{sample}"
     threads:
         get_threads(cluster_config, 'sentieon_dedup')

BALSAMIC/snakemake_rules/align/tga_sentieon_alignment.rule

@@ -12,9 +12,9 @@ rule sentieon_align_sort_umireads:
         Path(benchmark_dir, "sentieon_align_sort_{sample_type}_{sample}.tsv").as_posix()
     params:
         tmpdir = tempfile.mkdtemp(prefix=tmp_dir),
-        sentieon_install_dir= config["SENTIEON_INSTALL_DIR"],
-        sentieon_exec = config["SENTIEON_EXEC"],
-        sentieon_lic = config["SENTIEON_LICENSE"],
+        sentieon_install_dir = config_model.sentieon.sentieon_install_dir,
+        sentieon_exec = config_model.sentieon.sentieon_exec,
+        sentieon_lic = config_model.sentieon.sentieon_license,
         sample_id = "{sample}",
         sample_type = lambda wildcards: config_model.get_sample_type_by_name(wildcards.sample, uppercase=True),
         ip_bases = params.umicommon.align_intbases

BALSAMIC/snakemake_rules/align/wgs_bam_postprocess.rule

@@ -1,4 +1,47 @@
-"""Rules to mark duplicates and realign reads with Sentieon tools."""
+
+"""Rules to align, mark duplicates and realign reads with Sentieon tools."""
+
+rule sentieon_align_sort:
+    input:
+        ref = config["reference"]["reference_genome"],
+        fastq_r1 = Path(fastq_dir, "{fastq_pattern}_1.fp.fastq.gz").as_posix(),
+        fastq_r2 = Path(fastq_dir, "{fastq_pattern}_2.fp.fastq.gz").as_posix(),
+        refidx = expand(config["reference"]["reference_genome"] + ".{prefix}", prefix=["amb","ann","bwt","pac","sa"])
+    output:
+        bam_out = Path(bam_dir, "{sample}_align_sort_{fastq_pattern}.bam").as_posix()
+    benchmark:
+        Path(benchmark_dir, "sentieon_align_sort_{sample}_{fastq_pattern}.tsv").as_posix()
+    params:
+        tmpdir = tempfile.mkdtemp(prefix=tmp_dir),
+        header = params.common.align_header,
+        sentieon_exec = config_model.sentieon.sentieon_exec,
+        sentieon_lic = config_model.sentieon.sentieon_license,
+        sample_id = "{sample}",
+        sample_type = lambda wildcards: config_model.get_sample_type_by_name(wildcards.sample, uppercase=True),
+        fastq_pattern = "{fastq_pattern}"
+    threads:
+        get_threads(cluster_config, 'sentieon_align_sort')
+    message:
+        ("Align fastq reads using sentieon bwa-mem and sort reads using samtools for sample type: "
+        "{params.sample_type} : {params.sample_id}, {params.fastq_pattern}")
+    shell:
+        """
+mkdir -p {params.tmpdir};
+export TMPDIR={params.tmpdir};
+export SENTIEON_TMPDIR={params.tmpdir};
+export SENTIEON_LICENSE={params.sentieon_lic};
+
+{params.sentieon_exec} bwa mem -M \
+-R '@RG\\tID:{wildcards.fastq_pattern}\\tSM:{params.sample_type}\\tPL:ILLUMINA' \
+-t {threads} \
+-K 50000000 \
+{input.ref} {input.fastq_r1} {input.fastq_r2} \
+| {params.sentieon_exec} util sort \
+-o {output.bam_out} \
+-t {threads} \
+--block_size 3G \
+--sam2bam -i -;
+        """
 
 rule sentieon_dedup:
     input:
@@ -11,8 +54,8 @@ rule sentieon_dedup:
         Path(benchmark_dir, "sentieon_dedup_{sample_type}.{sample}.tsv").as_posix()
     params:
         tmpdir = tempfile.mkdtemp(prefix=tmp_dir),
-        sentieon_exec = config["SENTIEON_EXEC"],
-        sentieon_lic = config["SENTIEON_LICENSE"],
+        sentieon_exec = config_model.sentieon.sentieon_exec,
+        sentieon_lic = config_model.sentieon.sentieon_license,
         sample_id = "{sample}"
     threads:
         get_threads(cluster_config, 'sentieon_dedup')
@@ -55,8 +98,8 @@ rule sentieon_realign:
         Path(benchmark_dir, "sentieon_realign_{sample_type}.{sample}.tsv").as_posix()
     params:
         tmpdir = tempfile.mkdtemp(prefix=tmp_dir),
-         sentieon_exec = config["SENTIEON_EXEC"],
-        sentieon_lic = config["SENTIEON_LICENSE"],
+        sentieon_exec = config_model.sentieon.sentieon_exec,
+        sentieon_lic = config_model.sentieon.sentieon_license,
         sample_id = "{sample}"
     threads:
         get_threads(cluster_config, 'sentieon_realign')

BALSAMIC/snakemake_rules/align/wgs_sentieon_alignment.rule

@@ -12,8 +12,8 @@ rule sentieon_align_sort:
         Path(benchmark_dir, "sentieon_align_sort_{sample}_{fastq_pattern}.tsv").as_posix()
     params:
         tmpdir = tempfile.mkdtemp(prefix=tmp_dir),
-        sentieon_exec = config["SENTIEON_EXEC"],
-        sentieon_lic = config["SENTIEON_LICENSE"],
+        sentieon_exec = config_model.sentieon.sentieon_exec,
+        sentieon_lic = config_model.sentieon.sentieon_license,
         sample_id = "{sample}",
         sample_type = lambda wildcards: config_model.get_sample_type_by_name(wildcards.sample, uppercase=True),
         fastq_pattern = "{fastq_pattern}"

BALSAMIC/snakemake_rules/annotation/msi_tumor_normal.rule

@@ -47,7 +47,7 @@ rule msisensorpro_msi_tumor_normal:
     """
 msisensor-pro msi -b {threads} -z 1 -d {input.msi_list} -t {input.bamT} -n {input.bamN} -o  {params.tmpdir}/msi_{params.case_id};
 
-cp {params.tmpdir}/msi_{params.case_id} {output.msi_result}; 
+sed 's/\%/MSI/g' {params.tmpdir}/msi_{params.case_id} > {output.msi_result}; 
 
 rm -rf {params.tmpdir};
     """

BALSAMIC/snakemake_rules/quality_control/sentieon_qc_metrics.rule

@@ -24,8 +24,8 @@ if config["analysis"]["sequencing_type"] == 'wgs':
             min_base_qual = '10',
             gene_list = config["reference"]["refgene_txt"],
             cov_threshold = repeat("--cov_thresh", [50, 100, 150, 200, 250]),
-            sentieon_exec = config["SENTIEON_EXEC"],
-            sentieon_lic = config["SENTIEON_LICENSE"],
+            sentieon_exec = config_model.sentieon.sentieon_exec,
+            sentieon_lic = config_model.sentieon.sentieon_license,
             sample = '{sample}'
         threads:
             get_threads(cluster_config, 'sentieon_wgs_metrics')
@@ -72,8 +72,8 @@ if config["analysis"]["sequencing_type"] == 'wgs':
             Path(benchmark_dir, "sentieon_qc_metrics_{sample_type}.{sample}.tsv").as_posix()
         params:
             tmpdir = tempfile.mkdtemp(prefix=tmp_dir),
-            sentieon_exec = config["SENTIEON_EXEC"],
-            sentieon_lic = config["SENTIEON_LICENSE"],
+            sentieon_exec = config_model.sentieon.sentieon_exec,
+            sentieon_lic = config_model.sentieon.sentieon_license,
             sample_id = "{sample}",
             adapter = config["QC"]["adapter"]
         threads:
@@ -118,8 +118,8 @@ else:
             Path(benchmark_dir,"sentieon_qc_metrics_{sample_type}.{sample}.tsv").as_posix()
         params:
             tmpdir=tempfile.mkdtemp(prefix=tmp_dir),
-            sentieon_exec=config["SENTIEON_EXEC"],
-            sentieon_lic=config["SENTIEON_LICENSE"],
+            sentieon_exec=config_model.sentieon.sentieon_exec,
+            sentieon_lic=config_model.sentieon.sentieon_license,
             sample_id="{sample}",
             adapter=config["QC"]["adapter"]
         threads:
@@ -161,9 +161,12 @@ rule sentieon_plot_qc_metrics:
         Path(benchmark_dir, "sentieon_plot_qc_metrics_{sample_type}.{sample}.tsv").as_posix()
     params:
         tmpdir = tempfile.mkdtemp(prefix=tmp_dir),
-        sentieon_exec = config["SENTIEON_EXEC"],
-        sentieon_lic = config["SENTIEON_LICENSE"],
         sample_id = "{sample}",
+        min_base_qual = '10',
+        gene_list = config["reference"]["refgene_txt"],
+        cov_threshold = repeat("--cov_thresh", [50, 100, 150, 200, 250]),
+        sentieon_exec = config_model.sentieon.sentieon_exec,
+        sentieon_lic = config_model.sentieon.sentieon_license,
     threads:
         get_threads(cluster_config,'sentieon_plot_qc_metrics')
     message: