Merge pull request #6 from nf-core/alignment-subworkflow

Implement alignment subworkflow
nf-core · Apr 24, 2024 · 101e987 · 101e987
2 parents 48d34aa + e0da20a
commit 101e987
Show file tree

Hide file tree

Showing 51 changed files with 2,234 additions and 461 deletions.
diff --git a/conf/hmf_data.config b/conf/hmf_data.config
@@ -50,6 +50,7 @@ params {
             known_fusions                 = 'dna_pipeline/sv/known_fusions.37.bedpe'
             purple_germline_del           = 'dna_pipeline/copy_number/cohort_germline_del_freq.37.csv'
             segment_mappability           = 'dna_pipeline/variants/mappability_150.37.bed.gz'
+            unmap_regions                 = 'dna_pipeline/common/unmap_regions.37.tsv'
         }
         '38' {
             // AMBER
@@ -101,6 +102,7 @@ params {
             known_fusions                 = 'dna_pipeline/sv/known_fusions.38.bedpe'
             purple_germline_del           = 'dna_pipeline/copy_number/cohort_germline_del_freq.38.csv'
             segment_mappability           = 'dna_pipeline/variants/mappability_150.38.bed.gz'
+            unmap_regions                 = 'dna_pipeline/common/unmap_regions.38.tsv'
         }
     }
 }
diff --git a/conf/hmf_genomes.config b/conf/hmf_genomes.config
@@ -14,16 +14,22 @@ params {
             fai             = "https://pub-29f2e5b2b7384811bdbbcba44f8b5083.r2.dev/genomes/GRCh37_hmf/samtools_index/1.16/Homo_sapiens.GRCh37.GATK.illumina.fasta.fai"
             dict            = "https://pub-29f2e5b2b7384811bdbbcba44f8b5083.r2.dev/genomes/GRCh37_hmf/samtools_index/1.16/Homo_sapiens.GRCh37.GATK.illumina.fasta.dict"
             bwa_index       = "https://pub-29f2e5b2b7384811bdbbcba44f8b5083.r2.dev/genomes/GRCh37_hmf/bwa_index/0.7.17-r1188.tar.gz"
+            bwa_index_bseq  = "https://pub-29f2e5b2b7384811bdbbcba44f8b5083.r2.dev/genomes/GRCh37_hmf/bwa_index/2.2.1/Homo_sapiens.GRCh37.GATK.illumina.fasta.0123"
+            bwa_index_biidx = "https://pub-29f2e5b2b7384811bdbbcba44f8b5083.r2.dev/genomes/GRCh37_hmf/bwa_index/2.2.1/Homo_sapiens.GRCh37.GATK.illumina.fasta.bwt.2bit.64"
             bwa_index_image = "https://pub-29f2e5b2b7384811bdbbcba44f8b5083.r2.dev/genomes/GRCh37_hmf/bwa_index_image/0.7.17-r1188/Homo_sapiens.GRCh37.GATK.illumina.fasta.img"
             gridss_index    = "https://pub-29f2e5b2b7384811bdbbcba44f8b5083.r2.dev/genomes/GRCh37_hmf/gridss_index/2.13.2/Homo_sapiens.GRCh37.GATK.illumina.fasta.gridsscache"
+            star_index      = "https://pub-29f2e5b2b7384811bdbbcba44f8b5083.r2.dev/genomes/GRCh37_hmf/star_index/gencode_19/2.7.3a.tar.gz"
         }
         'GRCh38_hmf' {
             fasta           = "https://pub-29f2e5b2b7384811bdbbcba44f8b5083.r2.dev/genomes/GRCh38_hmf/GCA_000001405.15_GRCh38_no_alt_analysis_set.fna"
             fai             = "https://pub-29f2e5b2b7384811bdbbcba44f8b5083.r2.dev/genomes/GRCh38_hmf/samtools_index/1.16/GCA_000001405.15_GRCh38_no_alt_analysis_set.fna.fai"
             dict            = "https://pub-29f2e5b2b7384811bdbbcba44f8b5083.r2.dev/genomes/GRCh38_hmf/samtools_index/1.16/GCA_000001405.15_GRCh38_no_alt_analysis_set.fna.dict"
             bwa_index       = "https://pub-29f2e5b2b7384811bdbbcba44f8b5083.r2.dev/genomes/GRCh38_hmf/bwa_index/0.7.17-r1188.tar.gz"
+            bwa_index_bseq  = "https://pub-29f2e5b2b7384811bdbbcba44f8b5083.r2.dev/genomes/GRCh38_hmf/bwa_index/2.2.1/GCA_000001405.15_GRCh38_no_alt_analysis_set.fna.0123"
+            bwa_index_biidx = "https://pub-29f2e5b2b7384811bdbbcba44f8b5083.r2.dev/genomes/GRCh38_hmf/bwa_index/2.2.1/GCA_000001405.15_GRCh38_no_alt_analysis_set.fna.bwt.2bit.64"
             bwa_index_image = "https://pub-29f2e5b2b7384811bdbbcba44f8b5083.r2.dev/genomes/GRCh38_hmf/bwa_index_image/0.7.17-r1188/GCA_000001405.15_GRCh38_no_alt_analysis_set.fna.img"
             gridss_index    = "https://pub-29f2e5b2b7384811bdbbcba44f8b5083.r2.dev/genomes/GRCh38_hmf/gridss_index/2.13.2/GCA_000001405.15_GRCh38_no_alt_analysis_set.fna.gridsscache"
+            star_index      = "https://pub-29f2e5b2b7384811bdbbcba44f8b5083.r2.dev/genomes/GRCh38_hmf/star_index/gencode_38/2.7.3a.tar.gz"
         }
     }
 }
diff --git a/conf/modules.config b/conf/modules.config
@@ -12,6 +12,22 @@
 
 process {
 
+    withName: 'GATK4_MARKDUPLICATES' {
+        publishDir = [
+            path: { "${params.outdir}" },
+            mode: params.publish_dir_mode,
+            saveAs: { filename -> filename.equals('versions.yml') ? null : "${meta.key}/alignments/rna/${filename}" },
+        ]
+    }
+
+    withName: 'MARKDUPS' {
+        publishDir = [
+            path: { "${params.outdir}" },
+            mode: params.publish_dir_mode,
+            saveAs: { filename -> filename.equals('versions.yml') ? null : "${meta.key}/alignments/dna/${filename}" },
+        ]
+    }
+
     withName: 'AMBER' {
         publishDir = [
             path: { "${params.outdir}" },

diff --git a/lib/Constants.groovy b/lib/Constants.groovy
@@ -11,12 +11,12 @@ class Constants {
     static List PANELS_DEFINED     = ['tso500']
 
 
-    static String HMF_DATA_37_PATH = 'https://pub-29f2e5b2b7384811bdbbcba44f8b5083.r2.dev/hmf_reference_data/hmftools/5.34_37--0.tar.gz'
-    static String HMF_DATA_38_PATH = 'https://pub-29f2e5b2b7384811bdbbcba44f8b5083.r2.dev/hmf_reference_data/hmftools/5.34_38--0.tar.gz'
+    static String HMF_DATA_37_PATH = 'https://pub-29f2e5b2b7384811bdbbcba44f8b5083.r2.dev/hmf_reference_data/hmftools/5.34_37--2.tar.gz'
+    static String HMF_DATA_38_PATH = 'https://pub-29f2e5b2b7384811bdbbcba44f8b5083.r2.dev/hmf_reference_data/hmftools/5.34_38--2.tar.gz'
 
 
-    static String TSO500_PANEL_37_PATH = 'https://pub-29f2e5b2b7384811bdbbcba44f8b5083.r2.dev/hmf_reference_data/panels/tso500_5.34_37--0.tar.gz'
-    static String TSO500_PANEL_38_PATH = 'https://pub-29f2e5b2b7384811bdbbcba44f8b5083.r2.dev/hmf_reference_data/panels/tso500_5.34_38--0.tar.gz'
+    static String TSO500_PANEL_37_PATH = 'https://pub-29f2e5b2b7384811bdbbcba44f8b5083.r2.dev/hmf_reference_data/panels/tso500_5.34_37--1.tar.gz'
+    static String TSO500_PANEL_38_PATH = 'https://pub-29f2e5b2b7384811bdbbcba44f8b5083.r2.dev/hmf_reference_data/panels/tso500_5.34_38--1.tar.gz'
 
 
     static String VIRUSBREAKENDDB_PATH = 'https://pub-29f2e5b2b7384811bdbbcba44f8b5083.r2.dev/virusbreakend/virusbreakenddb_20210401.tar.gz'
@@ -34,6 +34,7 @@ class Constants {
     }
 
     static enum Process {
+        ALIGNMENT,
         AMBER,
         BAMTOOLS,
         CHORD,
@@ -45,6 +46,7 @@ class Constants {
         ISOFOX,
         LILAC,
         LINX,
+        MARKDUPS,
         ORANGE,
         PAVE,
         PURPLE,
@@ -56,7 +58,9 @@ class Constants {
     static enum FileType {
         // Generic
         BAM,
+        BAM_MARKDUPS,
         BAI,
+        FASTQ,
         // Process
         AMBER_DIR,
         BAMTOOLS,
@@ -97,11 +101,65 @@ class Constants {
         DNA_RNA,
     }
 
+    static enum InfoField {
+        CANCER_TYPE,
+        LANE,
+        LIBRARY_ID,
+    }
+
     static Map PLACEHOLDER_META = [meta_placeholder: null]
     static List PLACEHOLDER_OPTIONAL_CHANNEL = []
 
     static Map INPUT = [
 
+        BAM_DNA_TUMOR: [
+            FileType.BAM,
+            SampleType.TUMOR,
+            SequenceType.DNA,
+        ],
+
+        BAM_MARKDUPS_DNA_TUMOR: [
+            FileType.BAM_MARKDUPS,
+            SampleType.TUMOR,
+            SequenceType.DNA,
+        ],
+
+        BAM_DNA_NORMAL: [
+            FileType.BAM,
+            SampleType.NORMAL,
+            SequenceType.DNA,
+        ],
+
+        BAM_MARKDUPS_DNA_NORMAL: [
+            FileType.BAM_MARKDUPS,
+            SampleType.NORMAL,
+            SequenceType.DNA,
+        ],
+
+        BAM_RNA_TUMOR: [
+            FileType.BAM,
+            SampleType.TUMOR,
+            SequenceType.RNA,
+        ],
+
+        BAI_DNA_TUMOR: [
+            FileType.BAI,
+            SampleType.TUMOR,
+            SequenceType.DNA,
+        ],
+
+        BAI_DNA_NORMAL: [
+            FileType.BAI,
+            SampleType.NORMAL,
+            SequenceType.DNA,
+        ],
+
+        BAI_RNA_TUMOR: [
+            FileType.BAI,
+            SampleType.TUMOR,
+            SequenceType.RNA,
+        ],
+
         ISOFOX_DIR: [
             FileType.ISOFOX_DIR,
             SampleType.TUMOR,