Add HLA*LA support for HLA typing from BAM files

assets/software_meta.json

@@ -0,0 +1,11 @@
+{
+    "hlahd": {
+        "version": "1.7.1",
+        "software_md5": "44f0185f514f026fdd709fb34743980a"
+    },
+    "hlala": {
+        "graph": "PRG_MHC_GRCh38_withIMGT",
+        "graph_url": "https://www.well.ox.ac.uk/downloads/HLA-LA/graphs/PRG_MHC_GRCh38_withIMGT.tar.gz",
+        "graph_md5": "525a8aa0c7f357bf29fe2c75ef1d477d"
+    }
+}

conf/modules.config

@@ -20,18 +20,20 @@ process {
 
     withName: CHECK_PAIRED {
         ext.args = ''
+        publishDir = [ enabled: false ]
     }
 
     withName: FASTQC {
         ext.args = '--quiet'
     }
 
+    withName: YARA_INDEX {
+        publishDir = [ enabled: false ]
+    }
+
     withName: YARA_MAPPER {
         ext.args   = '-e 3'
-        publishDir = [
-            path: { "${params.outdir}/yara/${meta.id}" },
-            mode: params.publish_dir_mode
-        ]
+        publishDir = [ enabled: false ]
     }
 
     withName: OPTITYPE {
@@ -44,6 +46,10 @@ process {
         ext.args2 = "solver=${params.solver}"
     }
 
+    withName: 'HLAHD_INSTALL' {
+        publishDir = [ enabled: false ]
+    }
+
     withName: HLAHD {
         ext.args = '-m 50'
         publishDir = [
@@ -53,6 +59,31 @@ process {
         ]
     }
 
+    withName: 'HLALA_TYPING' {
+        ext.args = '--graph PRG_MHC_GRCh38_withIMGT'
+        publishDir = [
+            path: { "${params.outdir}/hlala/${meta.id}" },
+            mode: params.publish_dir_mode,
+            saveAs: { filename -> filename.equals('versions.yml') ? null : filename },
+        ]
+    }
+
+    withName: 'SAMTOOLS_VIEW' {
+        ext.args = '--output-fmt bam'  // Ensure BGZF-compressed BAM output
+        ext.prefix = { "${meta.id}.hlala" }
+        publishDir = [
+            enabled: false
+        ]
+    }
+
+    withName: 'HLALA_DOWNLOAD' {
+        publishDir = [ enabled: false ]
+    }
+
+    withName: 'SAMTOOLS_COLLATEFASTQ' {
+        publishDir = [ enabled: false ]
+    }
+
     withName: CAT_FASTQ {
         publishDir = [
             path: { params.save_merged_fastq ? "${params.outdir}/fastq" : params.outdir },

conf/test_hlala.config

@@ -0,0 +1,33 @@
+/*
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+    Nextflow config file for running HLA*LA tests
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+    Defines bundled input files and everything required to run a fast and simple test.
+
+    Use as follows:
+        nextflow run nf-core/hlatyping -profile test_hlala,<docker/singularity> --outdir <OUTDIR>
+
+    Note: HLA*LA requires a pre-built graph directory. If hlala_graph_dir is not provided,
+    the tool will attempt to download and build it automatically (requires significant
+    time and resources on first run).
+----------------------------------------------------------------------------------------
+ */
+
+process {
+    resourceLimits = [
+        cpus: 4,
+        memory: '15.GB',
+        time: '6.h'
+    ]
+}
+
+params {
+    config_profile_name        = 'HLA*LA Test Profile'
+    config_profile_description = 'Test dataset with BAM files using HLA*LA'
+
+    // Input data - HLA*LA requires pre-aligned BAM files
+    input = params.pipelines_testdata_base_path + 'hlatyping/samplesheets/samplesheet_bam.csv'
+    tools = 'hlala'
+
+    // HLA*LA graph directory - if not set, graph will be downloaded automatically
+}

conf/test_optitype_hlala.config

@@ -0,0 +1,32 @@
+/*
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+    Nextflow config file for running OptiType + HLA*LA tests
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+    Defines bundled input files and everything required to run a fast and simple test.
+
+    Use as follows:
+        nextflow run nf-core/hlatyping -profile test_optitype_hlala,<docker/singularity> --outdir <OUTDIR>
+----------------------------------------------------------------------------------------
+ */
+
+process {
+    resourceLimits = [
+        cpus: 4,
+        memory: '15.GB',
+        time: '1.h'
+    ]
+}
+
+params {
+    config_profile_name        = 'OptiType + HLA*LA Test Profile'
+    config_profile_description = 'Test dataset with BAM files using both OptiType and HLA*LA'
+
+    // Input data - HLA*LA requires pre-aligned BAM files
+    input           = params.pipelines_testdata_base_path + 'hlatyping/samplesheets/samplesheet_bam.csv'
+    // Parent directory containing the graph (e.g., graphs/PRG_MHC_GRCh38_withIMGT/)
+    hlala_graph_dir = "${projectDir}/graphs"
+    tools           = 'optitype,hlala'
+
+    // OptiType parameters
+    solver = 'glpk'
+}

modules.json

@@ -20,6 +20,16 @@
                         "git_sha": "41dfa3f7c0ffabb96a6a813fe321c6d1cc5b6e46",
                         "installed_by": ["modules"]
                     },
+                    "hlala/preparegraph": {
+                        "branch": "master",
+                        "git_sha": "5e748ff2b0f990949081c9e49792622eb3fe9ee9",
+                        "installed_by": ["modules"]
+                    },
+                    "hlala/typing": {
+                        "branch": "master",
+                        "git_sha": "e753770db613ce014b3c4bc94f6cba443427b726",
+                        "installed_by": ["modules"]
+                    },
                     "multiqc": {
                         "branch": "master",
                         "git_sha": "af27af1be706e6a2bb8fe454175b0cdf77f47b49",

modules/local/hlala/download/main.nf

@@ -0,0 +1,60 @@
+process HLALA_DOWNLOAD {
+    tag "$graph_name"
+    label 'process_single'
+
+    conda "conda-forge::wget=1.21.4 conda-forge::coreutils=9.5"
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        'https://depot.galaxyproject.org/singularity/wget:1.21.4' :
+        'biocontainers/wget:1.21.4' }"
+
+    input:
+    tuple val(graph_name), val(graph_url), val(graph_md5), path(graph_tarball)
+
+    output:
+    path "graphs", emit: graph
+    path "versions.yml", emit: versions
+
+    when:
+    task.ext.when == null || task.ext.when
+
+    script:
+    def use_local = graph_tarball.name != 'NO_FILE'
+    """
+    if [ "${use_local}" = "true" ]; then
+        # Use provided tarball
+        TARBALL="${graph_tarball}"
+    else
+        # Download graph
+        wget -q "${graph_url}" -O ${graph_name}.tar.gz
+        TARBALL="${graph_name}.tar.gz"
+    fi
+
+    # Verify MD5 checksum
+    checksum="\$(md5sum "\$TARBALL" | cut -f1 -d' ')"
+    if [ "\$checksum" != "${graph_md5}" ]; then
+        echo "Checksum error: expected ${graph_md5}, got \$checksum" >&2
+        exit 2
+    fi
+
+    # Extract into parent directory structure for HLA*LA --customGraphDir
+    mkdir -p graphs
+    tar -xzf "\$TARBALL" -C graphs
+    [ "${use_local}" = "false" ] && rm "\$TARBALL"
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        wget: \$(wget --version 2>&1 | head -1 | sed 's/GNU Wget //;s/ .*//')
+    END_VERSIONS
+    """
+
+    stub:
+    """
+    mkdir -p graphs/${graph_name}
+    touch graphs/${graph_name}/serializedGRAPH
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        wget: \$(wget --version 2>&1 | head -1 | sed 's/GNU Wget //;s/ .*//')
+    END_VERSIONS
+    """
+}