Sfitz input vcfs

config/custom_schema_types.config

@@ -2,9 +2,13 @@
 * This custom schema namespace implements a custom type for checking input BAMs as a list of Maps.
 */
 custom_schema_types {
-    allowed_sample_types = [
+    allowed_input_types = [
         'normal',
-        'tumor'
+        'tumor',
+        'muse',
+        'mutect2',
+        'somaticsniper',
+        'strelka2'
     ]
     allowed_resource_types = [
         'memory',
@@ -14,7 +18,7 @@ custom_schema_types {
     /**
     * Check that input types are in allowed list
     */
-    check_sample_type_keys = { List given, String name, List choices=custom_schema_types.allowed_sample_types ->
+    check_input_type_keys = { List given, String name, List choices=custom_schema_types.allowed_input_types ->
         for (elem in given) {
             if (!(elem in choices)) {
                 throw new Exception("Invalid paramter ${name}. Valid types: ${choices}.")
@@ -80,7 +84,7 @@ custom_schema_types {
         // Check parameters keys
         custom_schema_types.check_if_namespace(options[name], name)
         def given_keys = options[name].keySet() as ArrayList
-        custom_schema_types.check_sample_type_keys(given_keys, name)
+        custom_schema_types.check_input_type_keys(given_keys, name)
 
         options[name].each { entry ->
             def entry_as_map = [:]
@@ -98,7 +102,7 @@ custom_schema_types {
         if (given_keys.size() <= 0) {
             return
         }
-        custom_schema_types.check_sample_type_keys(given_keys, name, custom_schema_types.allowed_resource_types)
+        custom_schema_types.check_input_type_keys(given_keys, name, custom_schema_types.allowed_resource_types)
 
         options[name].each { entry ->
             def entry_as_map = [:]
@@ -120,6 +124,19 @@ custom_schema_types {
         }
     }
 
+    /**
+    * Check if proper VCF entry list
+    */
+    check_vcf_list = { Map options, String name, Map properties ->
+        custom_schema_types.check_if_list(options[name], name)
+        for (item in options[name]) {
+            custom_schema_types.check_if_namespace(item, name)
+            properties.elements.each { key, val ->
+                schema.validate_parameter(item, key, val)
+            }
+        }
+    }
+
     /**
     *   Check list of resource updates
     */
@@ -134,6 +151,7 @@ custom_schema_types {
     types = [
         'InputNamespace': custom_schema_types.check_input_namespace,
         'BAMEntryList': custom_schema_types.check_bam_list,
+        'VCFEntryList': custom_schema_types.check_vcf_list,
         'ResourceUpdateNamespace': custom_schema_types.check_resource_update_namespace,
         'ResourceUpdateList': custom_schema_types.check_resource_update_list
     ]

config/default.config

@@ -5,6 +5,7 @@ params {
     ucla_cds = true
 
     cache_intermediate_pipeline_steps = false
+    keep_input_prefix = false
 
     //max_number_of_parallel_jobs = 1
     max_cpus   = SysHelper.getAvailCpus()

config/methods.config

@@ -53,6 +53,18 @@ methods {
         }
     }
 
+    get_vcfs_to_process = {
+        params.tumor_id = methods.sanitize_string(params.input_tumor_id)
+        params.normal_id = methods.sanitize_string(params.input_normal_id)
+        params.samples_to_process = [] as Set
+        params.input.each { k, v ->
+            params.samples_to_process.add([
+                'path': v,
+                'algorithm': k]
+            )
+        }
+    }
+
     set_sample_params = {
         params.single_NT_paired = false
         def tumor_ids = params.samples_to_process.findAll { it['sample_type'] == 'tumor' }['id']
@@ -71,17 +83,6 @@ methods {
         }
     }
 
-    set_intersect_regions_params = {
-        if (params.containsKey("intersect_regions") && params.intersect_regions) {
-            params.intersect_regions_index = "${params.intersect_regions}.tbi"
-            params.use_intersect_regions = true
-        } else {
-            params.intersect_regions = "${params.work_dir}/NO_FILE.bed"
-            params.intersect_regions_index = "${params.work_dir}/NO_FILE.bed.tbi"
-            params.use_intersect_regions = false
-        }
-    }
-
     set_mutect2_params = {
         if (params.containsKey("germline_resource_gnomad_vcf") && params.germline_resource_gnomad_vcf) {
             params.germline = true
@@ -92,13 +93,33 @@ methods {
         params.germline_resource_gnomad_vcf_index = "${params.germline_resource_gnomad_vcf}.tbi"
     }
 
+    check_valid_algorithms = {
+        valid_algorithms = params.single_NT_paired ? ['somaticsniper', 'strelka2', 'mutect2', 'muse'] : ['mutect2']
+        for (algo in params.algorithm) {
+            if (!(algo in valid_algorithms)) {
+                throw new Exception("ERROR: params.algorithm ${params.algorithm} contains an invalid value. Valid algorithms for given inputs: ${valid_algorithms}")
+            }
+        }
+    }
+
     set_output_directory = {
         def tz = TimeZone.getTimeZone("UTC")
         def date = new Date().format("yyyyMMdd'T'HHmmss'Z'", tz)
         params.output_dir_base = "${params.output_dir}/${manifest.name}-${manifest.version}/${params.sample_id}"
         params.log_output_dir = "${params.output_dir_base}/log-${manifest.name}-${manifest.version}-${date}"
     }
 
+    set_intersect_regions_params = {
+        if (params.containsKey("intersect_regions") && params.intersect_regions) {
+            params.intersect_regions_index = "${params.intersect_regions}.tbi"
+            params.use_intersect_regions = true
+        } else {
+            params.intersect_regions = "${params.work_dir}/NO_FILE.bed"
+            params.intersect_regions_index = "${params.work_dir}/NO_FILE.bed.tbi"
+            params.use_intersect_regions = false
+        }
+    }
+
     set_pipeline_log = {
         trace.enabled = true
         trace.file = "${params.log_output_dir}/nextflow-log/trace.txt"
@@ -110,15 +131,6 @@ methods {
         report.file = "${params.log_output_dir}/nextflow-log/report.html"
     }
 
-    check_valid_algorithms = {
-        valid_algorithms = params.single_NT_paired ? ['somaticsniper', 'strelka2', 'mutect2', 'muse'] : ['mutect2']
-        for (algo in params.algorithm) {
-            if (!(algo in valid_algorithms)) {
-                throw new Exception("ERROR: params.algorithm ${params.algorithm} contains an invalid value. Valid algorithms for given inputs: ${valid_algorithms}")
-            }
-        }
-    }
-
     setup = {
         schema.load_custom_types("${projectDir}/config/custom_schema_types.config")
         schema.validate()
@@ -127,13 +139,22 @@ methods {
         methods.modify_base_allocations()
         retry.setup_retry()
         methods.set_env()
-        methods.get_ids_from_bams()
-        methods.set_sample_params()
-        methods.set_intersect_regions_params()
-        methods.set_mutect2_params()
+        if (params.input.containsKey('tumor')) {
+            params.input_type = 'bam'
+            methods.get_ids_from_bams()
+            methods.set_sample_params()
+            methods.set_mutect2_params()
+            methods.check_valid_algorithms()
+        } else if (params.input.containsKey('muse') || params.input.containsKey('somaticsniper') || params.input.containsKey('strelka2') || params.input.containsKey('mutect2')) {
+            params.input_type = 'vcf'
+            methods.get_vcfs_to_process()
+            params.sample_id = params.tumor_id
+        } else {
+            throw new Exception("ERROR: No input yamls detected.")
+        }
         methods.set_output_directory()
+        methods.set_intersect_regions_params()
         methods.set_pipeline_log()
-        methods.check_valid_algorithms()
         methods.setup_docker_cpus()
     }
 }

config/schema.yaml

@@ -112,11 +112,11 @@ base_resource_update:
 input:
   type: 'InputNamespace'
   required: true
-  help: 'Input samples'
+  help: 'Input to process'
   elements:
     tumor:
       type: 'BAMEntryList'
-      required: true
+      required: false
       help: 'Tumor id/path input'
       elements:
         BAM:
@@ -139,3 +139,23 @@ input:
           mode: 'r'
           required: true
           help: 'Absolute path to normal sample BAM files'
+    muse:
+      type: 'Path'
+      mode: 'r'
+      required: false
+      help: 'Absolute path to muse VCF file'
+    mutect2:
+      type: 'Path'
+      mode: 'r'
+      required: false
+      help: 'Absolute path to mutect2 VCF file'
+    somaticsniper:
+      type: 'Path'
+      mode: 'r'
+      required: false
+      help: 'Absolute path to somaticsniper VCF file'
+    strelka2:
+      type: 'Path'
+      mode: 'r'
+      required: false
+      help: 'Absolute path to strelka2 VCF file'

input/call-sSNV-template-VCF.yaml

@@ -0,0 +1,13 @@
+---
+patient_id: 'patient_id'
+tumor_id: 'tumor_id' # must match the tumor_id in the VCF files
+normal_id: 'normal_id' # must match the normal_id in the VCF files
+input:
+  muse:
+    - VCF: /path/to/muse.vcf.gz
+  mutect2:
+    - VCF: /path/to/mutect2.vcf.gz
+  somaticsniper:
+    - VCF: /path/to/somaticsniper.vcf.gz
+  strelka2:
+    - VCF: /path/to/strelka2.vcf.gz