feat: add msisensorpro TN

BALSAMIC/constants/cluster_analysis.json

@@ -415,5 +415,13 @@
 	"igh_dux4_detection": {
 		"time": "02:00:00",
 		"n": 1
+	},
+	"msisensorpro_scan_reference": {
+		"time": "04:00:00",
+		"n": 1
+	},
+	"msisensorpro_msi_tumor_normal": {
+		"time": "08:00:00",
+		"n": 24
 	}
 }

BALSAMIC/snakemake_rules/annotation/somatic_computations.rule

@@ -59,3 +59,54 @@ bcftools view --types snps,indels --apply-filters PASS {params.tmpdir}/temp.vcf.
 rm -r {params.tmpdir};
         """
 
+if config["analysis"]["analysis_type"] == "paired":
+  rule msisensorpro_scan_reference:
+    input:
+      fa = config["reference"]["reference_genome"],
+    output:
+      msi_scan =  f"{vcf_dir}CNV.somatic.{config['analysis']['case_id']}.msisensorpro.list"
+    benchmark:
+        Path(f"{benchmark_dir}msisensorpro_scan_reference_{config['analysis']['case_id']}.tsv").as_posix()
+    singularity:
+        Path(singularity_image, config["bioinfo_tools"].get("msisensorpro") + ".sif").as_posix()
+    threads:
+        get_threads(cluster_config, "msisensorpro_scan_reference")
+    params:
+        tmpdir = tempfile.mkdtemp(prefix=tmp_dir),
+        case_name = config["analysis"]["case_id"],
+    message:
+        "Scanning microsatellite sites msisensor-pro for {params.case_name}"
+    shell:
+      """
+msisensor-pro scan -d {input.fa} -o {output.msi_scan};
+
+rm -rf {params.tmpdir};
+      """
+
+
+  rule msisensorpro_msi_tumor_normal:
+    input:
+      msi_list=f"{vcf_dir}CNV.somatic.{config['analysis']['case_id']}.msisensorpro.list",
+      bamN= config_model.get_final_bam_name(bam_dir = bam_dir,sample_name=normal_sample),
+      bamT= config_model.get_final_bam_name(bam_dir = bam_dir,sample_name=tumor_sample),
+    output:
+      msi_result = f"{vcf_dir}CNV.somatic.{config['analysis']['case_id']}.msisensorpro.msi"
+    benchmark:
+      Path(f"{benchmark_dir}/msisensorpro_msi_tumor_normal_{config['analysis']['case_id']}.tsv").as_posix()
+    singularity:
+      Path(singularity_image,config["bioinfo_tools"].get("msisensorpro") + ".sif").as_posix()
+    threads:
+      get_threads(cluster_config,"msisensorpro_msi_tumor_normal")
+    params:
+      tmpdir=tempfile.mkdtemp(prefix=tmp_dir),
+      case_name=config["analysis"]["case_id"],
+    message:
+      "Analysing MSI for {params.case_name}"
+    shell:
+      """
+msisensor-pro msi -b {threads} -z 1 -d {input.msi_list} -t {input.bamT} -n {input.bamN} -o  {params.tmpdir}/msi_{params.case_name};
+
+cp {params.tmpdir}/msi_{params.case_name} {output.msi_result}; 
+
+rm -rf {params.tmpdir};
+      """

BALSAMIC/workflows/balsamic.smk

@@ -287,6 +287,7 @@ os.environ['TMPDIR'] = get_result_dir(config)
 cnv_report_paths = []
 if config["analysis"]["sequencing_type"] == "wgs":
     if config['analysis']['analysis_type'] == "paired":
+        cnv_report_paths.append(f"{vcf_dir}CNV.somatic.{config['analysis']['case_id']}.msisensorpro.msi")
         cnv_report_paths.append(f"{vcf_dir}CNV.somatic.{config['analysis']['case_id']}.ascat.samplestatistics.txt.pdf")
         cnv_report_paths.extend(expand(
             f"{vcf_dir}CNV.somatic.{config['analysis']['case_id']}.ascat.{{output_suffix}}.png.pdf",
@@ -298,6 +299,8 @@ if config["analysis"]["sequencing_type"] == "wgs":
             output_suffix=["circular", "scatter"]
         ))
 else:
+    if config["analysis"]["analysis_type"] == "paired":
+        cnv_report_paths.append(f"{vcf_dir}CNV.somatic.{config['analysis']['case_id']}.msisensorpro.msi")
     cnv_report_paths.extend(expand(f"{cnv_dir}tumor.merged-{{plot}}.pdf",plot=["diagram", "scatter"]))
     cnv_report_paths.append(f"{cnv_dir}CNV.somatic.{config['analysis']['case_id']}.purecn.purity.csv.pdf")
 

CHANGELOG.rst

@@ -5,6 +5,8 @@
 Added:
 ^^^^^^
 * MSIsensor-pro container https://github.com/Clinical-Genomics/BALSAMIC/pull/1444
+* MSI anslysis to the tumor-normal workflow https://github.com/Clinical-Genomics/BALSAMIC/pull/1454 
+
 
 Changed:
 ^^^^^^^^