Sex prediction

assets/multiqc_config.yml

@@ -12,3 +12,31 @@ report_section_order:
 export_plots: true
 
 disable_version_detection: true
+
+table_cond_formatting_rules:
+  match_certainty:
+    pass:
+      - s_eq: 100
+    warn:
+      - lt: 99
+    fail:
+      - lt: 50
+
+table_columns_name:
+  sex_prediction:
+    match_certainty: "Certainty of sex match"
+    expected_sex: "Expected Sex"
+    calculated_sex: "Calculated Sex"
+    calculated_sex_yx: "Calculated sex (XY method)"
+    calculated_sex_sry: "Calculated sex (SRY method)"
+    calculated_sex_hetx: "Calculated sex (HETX method)"
+    ratio_yx: "# reads in chrY / # reads in chrX (XY)"
+    het_fraction: "Fraction of heterozygous SNPs in chrX (HETX)"
+    coverage_sry: "Coverage of SRY in chrY (SRY)"
+    xy_ratio: "X/Y ratio"
+
+custom_plot_config:
+  sex_prediction:
+    defaultsort:
+      - column: "Certainty of sex match"
+        direction: asc

assets/schema_input.json

@@ -57,6 +57,13 @@
                 "format": "file-path",
                 "exists": true,
                 "errorMessage": "snp BAM/CRAM index file must be provided, cannot contain spaces and must have extension '.bai' or '.csi'"
+            },
+            "sex": {
+                "type": "string",
+                "enum": ["M", "F", "U"],
+                "default": "U",
+                "meta": "sex",
+                "errorMessage": "The value should be one of 'M' (male), 'F' (female) or 'U' (unknown)"
             }
         },
         "uniqueEntries": ["sample"],

conf/WGS.config

@@ -0,0 +1,10 @@
+/*
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+    Nextflow config file for running the pipeline on WGS data
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+*/
+
+params {
+    min_xy_female = 4.5
+    max_xy_male = 2
+}

conf/modules.config

@@ -43,7 +43,22 @@ process {
         ].join(" ").trim()}
     }
 
-    withName: 'MULTIQC' {
+    withName: "NGSBITS_SAMPLEGENDER_XY" {
+        ext.prefix = { "${meta.id}.xy" }
+        ext.args = "-max_female ${params.max_yx_female} -min_male ${params.min_yx_male}"
+    }
+
+    withName: "NGSBITS_SAMPLEGENDER_SRY" {
+        ext.prefix = { "${meta.id}.sry" }
+        ext.args = "-sry_cov ${params.min_sry_cov_male}"
+    }
+
+    withName: "NGSBITS_SAMPLEGENDER_HETX" {
+        ext.prefix = { "${meta.id}.hetx" }
+        ext.args = "-min_female ${params.min_hetx_female} -max_male ${params.max_hetx_male}"
+    }
+
+    withName: 'MULTIQC_POOLS' {
         ext.args   = { params.multiqc_title ? "--title \"$params.multiqc_title_$meta.id\"" : "--title \"$meta.id\"" }
         publishDir = [
             path: { "${params.outdir}/multiqc" },
@@ -52,4 +67,13 @@ process {
         ]
     }
 
+    withName: 'MULTIQC_MAIN' {
+        ext.args   = { params.multiqc_title ? "--title \"$params.multiqc_title\"" : '' }
+        publishDir = [
+            path: { "${params.outdir}/multiqc" },
+            mode: params.publish_dir_mode,
+            saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
+        ]
+    }
+
 }

main.nf

@@ -63,6 +63,7 @@ workflow {
             file(params.haplotype_map, checkIfExists: true)
         ]),
         params.outdir,
+        params.skip_steps,
         params.multiqc_config,
         params.multiqc_logo,
         params.multiqc_methods_description

modules.json

@@ -21,6 +21,11 @@
                         "installed_by": ["modules"],
                         "patch": "modules/nf-core/multiqc/multiqc.diff"
                     },
+                    "ngsbits/samplegender": {
+                        "branch": "master",
+                        "git_sha": "9e13262b5632b7a1e8247de7ad62fca0e9bbc953",
+                        "installed_by": ["modules"]
+                    },
                     "picard/crosscheckfingerprints": {
                         "branch": "master",
                         "git_sha": "49f4e50534fe4b64101e62ea41d5dc43b1324358",

nextflow.config

@@ -10,13 +10,23 @@
 params {
 
     // Input options
-    input                      = null
+    input                       = null
 
     // References
-    fasta                      = null
-    fai                        = null
-    bwa_index                  = null
-    haplotype_map              = null
+    fasta                       = null
+    fai                         = null
+    bwa_index                   = null
+    haplotype_map               = null
+
+    // Pipeline options
+    skip_steps                  = ""
+
+    // Sex prediction
+    max_yx_female               = 0.06
+    min_yx_male                 = 0.09
+    min_sry_cov_male            = 20
+    min_hetx_female             = 0.25
+    max_hetx_male               = 0.05
 
     // MultiQC options
     multiqc_config              = null