Parabricks starfusion

.github/workflows/nf-test-gpu.yml

@@ -91,7 +91,7 @@ jobs:
           echo ${{ steps.set-shards.outputs.total_shards }}
 
   nf-test-gpu:
-    runs-on: "runs-on=${{ github.run_id }}/family=g4dn.xlarge/image=ubuntu24-gpu-x64"
+    runs-on: "runs-on=${{ github.run_id }}/family=g4dn.8xlarge/image=ubuntu24-gpu-x64"
     name: "GPU Test | ${{ matrix.profile }} | ${{ matrix.shard }}"
     needs: [nf-test-changes]
     if: ${{ needs.nf-test-changes.outputs.total_shards != '0' }}

modules/nf-core/parabricks/README.md

@@ -1,18 +1,20 @@
-# NVIDIA Clara Parabricks modules
+# NVIDIA Parabricks Modules
 
-These nf-core modules implement functionality of the NVIDIA Clara Parabricks programs for GPU-accelerated genomics tasks. Parabricks covers the functionality of alignment (replicating `bwa mem`), variant calling (replicating `gatk4` and `deepvariant`), and other common genomics tasks. Please see the [documentation](https://docs.nvidia.com/clara/parabricks/4.0.1/index.html) for additional details.
+These nf-core modules implement functionality of the NVIDIA Parabricks software for GPU-accelerated genomics tasks. Parabricks covers the functionality of alignment (Ex. `bwa mem`, `bwa meth` and `STAR`), variant and fusion calling (Ex. `deepvariant`, `mutectcaller`, and `starfusion`), and other common genomics tasks. Please see the [documentation](https://docs.nvidia.com/clara/parabricks/latest/index.html) for additional details.
 
 ## General considerations
 
-Parabricks is available only through a docker container: `nvcr.io/nvidia/clara/clara-parabricks:4.5.1-1`. The main entrypoint to the Parabricks is the command line program `pbrun`, which calls several sub-programs within the container. The Prabricks authors combined several common patterns into a single tool, such as `fq2bam` performing alignment, sorting, mark duplicates, and base quality score recalibration, all within a single command. Generally, the tools can be used for only a subset of the entire functionality as well.
+Parabricks is available only through a docker container: `nvcr.io/nvidia/clara/clara-parabricks:latest`. The main entrypoint to the Parabricks is the command line program `pbrun`, which calls several sub-programs within the container. The Parabricks authors sometimes combine several common tasks into a single tool, such as `fq2bam` performing bwa alignment, sorting, mark duplicates, and base quality score recalibration, all within a single command. Generally, the tools can be used for only a subset of the entire functionality as well.
 
-Parabricks tools must be run with at least one NVIDIA GPU with >16GB vRAM, and a usually require a large amount of resources (at least 8 threads and 30GB RAM). Please see the [resource requirements](https://docs.nvidia.com/clara/parabricks/4.0.1/GettingStarted.html) for more information. It is recommended to use fast local storage whenever possible to ensure the system is not bottlenecked by I/O.
+## Hardware Requirements
+
+Parabricks tools must be run with at least one NVIDIA GPU with at least 16GB vRAM, and a usually require a large amount of resources (at least 8 threads and 30GB RAM). Please see the [installation requirements](https://docs.nvidia.com/clara/parabricks/latest/gettingstarted/installationrequirements.html) for more information. It is recommended to use fast local storage whenever possible to ensure the system is not bottlenecked by I/O.
 
 To give docker or singularity access to GPUs present on the host system, add this line to the configuration file: `docker.runOptions = "--gpus all"` or `singularity.runOptions = "--nv"`.
 
 ## License
 
-As of version 4.0, Parabricks is available for general use. The license from NVIDIA states:
+As of version 4.0, Parabricks is available for free for general use. The license from NVIDIA states:
 
 ```
 A license is no longer required to use Clara Parabricks. The container works out of the box once downloaded.
@@ -26,24 +28,44 @@ Clara Parabricks is free for
 Users who would like to have Enterprise Support for Clara Parabricks can purchase NVIDIA AI Enterprise licenses, which provides full-stack support. To learn more about NVIDIA AI Enterprise, please visit https://www.nvidia.com/en-us/data-center/products/ai-enterprise/.
 ```
 
-## Specific tools
+## Notes on Specific tools
 
-### parabricks/fq2bam
+### fq2bam
 
-`fq2bam` performs alignment, sorting, (optional) marking of duplicates, and (optional) base quality score recalibration (BQSR). There is no option to control the number of threads used with this tool - all available threads on the system are used by default.
+The `fq2bam` module performs alignment, sorting, (optional) marking of duplicates, and (optional) base quality score recalibration (BQSR). There is no option to control the number of threads used with this tool - all available threads on the system are used by default.
 
 Alignment and coordinate sorting are always performed. Duplicate marking can be performed by passing the option `markdups=true`. Duplicate marking and BQSR can be performed by passing the options `markdups=true` and `known_sites=$KNOWN_SITES_FILE`.
 
 Please see the `fq2bam/meta.yml` file for a detailed list of required and optional inputs and outputs.
 
 For additional considerations, including information about how readgroups are added to the resulting bam files, see the [tool documentation](https://docs.nvidia.com/clara/parabricks/latest/Documentation/ToolDocs/man_fq2bam.html).
 
-## parabricks/stargenomegenerate
+## rnafq2bam
+
+The `rnafq2bam` module is based on STAR version `2.7.2a`. Therefore the genome lib directory required as input for this module must also be generated using this version of STAR. For convenience, a module with this version of STAR is included in this directory under `parabricks/stargenomegenerate`.
+
+## starfusion
 
-The `parabricks/stargenomegenerate` module is near identical to the existing `star/genomegenerate` module, however it runs on with older version of STAR (2.7.2a) that is required for Parabricks compatibility. This module does not exist in any previous versions of the `nf-core/modules` and therefore must be included here. In the future, it's possible that Parabricks will update to a newer version of STAR and this accessory module may become obselete, but for now it is required pre-processing if the Genome Lib Dir has not already been generated with this vesion of STAR.
+The `starfusion` module is based on starfusion 1.7.0. Therefore the genome lib directory required as input for this module must also be generated using this version of starfusion. For convenience, a module with this version of starfusion is included in this directory under `parabricks/starfusion_build`.
 
-## Compatible with
+## compatible_with.yaml
 
-Is added as optional output to the stub section to make the compatible CPU version available to the end user. This section is not given for the subtools `applybqsr`, `fq2bammeth`, `genotypegvcf`, or `rnafq2bam`.
+The `compatible_with.yaml` is provided as an optional output to the stub section to make the compatible CPU version available to the end user. This section is not given for the subtools `applybqsr`, `fq2bammeth`, `genotypegvcf`, `rnafq2bam`, or `starfusion`.
 
 For the full list of compatible versions, check the [Parabricks documentation](https://docs.nvidia.com/clara/parabricks/latest/documentation/tooldocs/outputaccuracyandcompatiblecpusoftwareversions.html#).
+
+## Notes on Testing
+
+The Parabricks `starfusion` module requires testing on a `g4dn.8xlarge` instead of the default `g4dn.xlarge` due to higher system memory requirements. In [nf-test-gpu.yml](../../../.github/workflows/nf-test-gpu.yml), update the following line: 
+
+```
+nf-test-gpu:
+    runs-on: "runs-on=${{ github.run_id }}/family=g4dn.8xlarge/image=ubuntu24-gpu-x64"
+```
+
+Change it back before merging into master:
+
+```
+nf-test-gpu:
+    runs-on: "runs-on=${{ github.run_id }}/family=g4dn.xlarge/image=ubuntu24-gpu-x64"
+```

modules/nf-core/parabricks/rnafq2bam/main.nf

@@ -5,18 +5,21 @@ process PARABRICKS_RNAFQ2BAM {
     // needed by the module to work properly can be removed when fixed upstream - see: https://github.com/nf-core/modules/issues/7226
     stageInMode 'copy'
 
-    container "nvcr.io/nvidia/clara/clara-parabricks:4.5.1-1"
+    container "nvcr.io/nvidia/clara/clara-parabricks:4.6.0-1"
 
     input:
-    tuple val(meta), path(reads)
+    tuple val(meta),  path(reads)
     tuple val(meta1), path(fasta)
     tuple val(meta2), path(index)
     tuple val(meta3), path(genome_lib_dir)
 
     output:
-    tuple val(meta), path("*.bam"), emit: bam
-    tuple val(meta), path("*.bai"), emit: bai
-    path "versions.yml",            emit: versions
+    tuple val(meta), path("*.bam"),                                 emit: bam
+    tuple val(meta), path("*.bai"),                                 emit: bai
+    tuple val(meta), path("Chimeric.out.junction"),                 emit: junction,             optional: true
+    tuple val(meta), path("*_qc_metrics"),                          emit: qc_metrics,           optional:true
+    tuple val(meta), path("*.duplicate-metrics.txt"),               emit: duplicate_metrics,    optional:true
+    path "versions.yml",                                            emit: versions
 
     when:
     task.ext.when == null || task.ext.when
@@ -46,6 +49,10 @@ process PARABRICKS_RNAFQ2BAM {
         ${num_gpus} \\
         ${args}
 
+    if [[ "${args}" == *"--out-chim-type"* ]]; then
+        mv ${prefix}/Chimeric.out.junction .
+    fi
+
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":
             pbrun: \$(echo \$(pbrun version 2>&1) | sed 's/^Please.* //' )
@@ -59,11 +66,13 @@ process PARABRICKS_RNAFQ2BAM {
     }
     def args = task.ext.args ?: ''
     def prefix = task.ext.prefix ?: "${meta.id}"
+    def chimeric_output = args.contains("--out-chim-type") ? "touch Chimeric.out.junction" : ""
     def qc_metrics_output = args.contains("--out-qc-metrics-dir") ? "mkdir ${prefix}_qc_metrics" : ""
     def duplicate_metrics_output = args.contains("--out-duplicate-metrics") ? "touch ${prefix}.duplicate-metrics.txt" : ""
     """
     touch ${prefix}.bam
     touch ${prefix}.bam.bai
+    ${chimeric_output}
     ${qc_metrics_output}
     ${duplicate_metrics_output}
 

modules/nf-core/parabricks/rnafq2bam/meta.yml

@@ -81,6 +81,39 @@ output:
           pattern: "*.bai"
           ontologies:
             - edam: http://edamontology.org/format_3327 # BAI
+  junction:
+    - - meta:
+          type: map
+          description: |
+            Groovy Map containing sample information
+            e.g. [ id:'test', single_end:false ]
+      - "Chimeric.out.junction":
+          type: file
+          description: Chimeric junction output file
+          pattern: "Chimeric.out.junction"
+          ontologies: []
+  qc_metrics:
+    - - meta:
+          type: map
+          description: |
+            Groovy Map containing sample information
+            e.g. [ id:'test', single_end:false ]
+      - "*_qc_metrics":
+          type: directory
+          description: (optional) optional directory of qc metrics
+          pattern: "*_qc_metrics"
+  duplicate_metrics:
+    - - meta:
+          type: map
+          description: |
+            Groovy Map containing sample information
+            e.g. [ id:'test', single_end:false ]
+      - "*.duplicate-metrics.txt":
+          type: file
+          description: (optional) metrics calculated from marking duplicates in the
+            bam file
+          pattern: "*.duplicate-metrics.txt"
+          ontologies: []
   versions:
     - versions.yml:
         type: file

modules/nf-core/parabricks/rnafq2bam/tests/main.nf.test

@@ -18,14 +18,14 @@ nextflow_process {
             script "../../stargenomegenerate/main.nf"
             process {
                 """
-                input[0] = Channel.of([
-                    [ id:'test_fasta' ],
-                    [ file(params.modules_testdata_base_path + 'genomics/homo_sapiens/genome/genome.fasta', checkIfExists: true) ]
-                ])
-                input[1] = Channel.of([
-                    [ id:'test_gtf' ],
-                    [ file(params.modules_testdata_base_path + 'genomics/homo_sapiens/genome/genome.gtf', checkIfExists: true) ]
-                ])
+                input[0] = [
+                    [ id:'minigenome_fasta' ],
+                    file(params.modules_testdata_base_path + 'genomics/homo_sapiens/genome/minigenome.fa')
+                ]
+                input[1] = [
+                    [ id:'minigenome_gtf' ],
+                    file(params.modules_testdata_base_path + 'genomics/homo_sapiens/genome/minigenome.gtf')
+                ]
                 """
             }
         }
@@ -49,7 +49,7 @@ nextflow_process {
 
         when {
             params {
-                module_args = '--low-memory --read-files-command zcat'
+                module_args = '--low-memory --read-files-command zcat --x3'
                 // Ref: https://forums.developer.nvidia.com/t/problem-with-gpu/256825/6
                 // Parabricks’s rnafq2bam requires 24GB of memory.
                 // Using --low-memory for testing
@@ -58,11 +58,11 @@ nextflow_process {
                 """
                 input[0] = Channel.of([
                     [ id:'test', single_end:true ], // meta map
-                    [ file(params.modules_testdata_base_path + 'genomics/homo_sapiens/illumina/fastq/test_rnaseq_1.fastq.gz', checkIfExists: true) ]
+                    [ file(params.modules_testdata_base_path + 'genomics/homo_sapiens/genome/test_starfusion_rnaseq_1.fastq.gz', checkIfExists: true) ]
                 ])
                 input[1] = Channel.of([
                     [ id:'test' ], // meta map
-                    file(params.modules_testdata_base_path + 'genomics/homo_sapiens/genome/genome.fasta', checkIfExists: true)
+                    file(params.modules_testdata_base_path + 'genomics/homo_sapiens/genome/minigenome.gtf', checkIfExists: true)
                 ])
 				input[2] = BWA_INDEX.out.index
                 input[3] = PARABRICKS_STARGENOMEGENERATE.out.index
@@ -96,11 +96,88 @@ nextflow_process {
                 """
                 input[0] = Channel.of([
                     [ id:'test', single_end:true ], // meta map
-                    [ file(params.modules_testdata_base_path + 'genomics/homo_sapiens/illumina/fastq/test_rnaseq_1.fastq.gz', checkIfExists: true) ]
+                    [ file(params.modules_testdata_base_path + 'genomics/homo_sapiens/genome/test_starfusion_rnaseq_1.fastq.gz', checkIfExists: true) ]
                 ])
                 input[1] = Channel.of([
                     [ id:'test' ], // meta map
-                    file(params.modules_testdata_base_path + 'genomics/homo_sapiens/genome/genome.fasta', checkIfExists: true)
+                    file(params.modules_testdata_base_path + 'genomics/homo_sapiens/genome/minigenome.gtf', checkIfExists: true)
+                ])
+				input[2] = BWA_INDEX.out.index
+                input[3] = PARABRICKS_STARGENOMEGENERATE.out.index
+                """
+            }
+        }
+
+        then {
+            assertAll(
+                { assert process.success },
+                { assert snapshot(
+                    process.out,
+                    path(process.out.versions[0]).yaml
+                ).match() }
+            )
+        }
+    }
+
+    test("homo_sapiens_chimeric") {
+
+        config "./nextflow.config"
+
+        when {
+            params {
+                module_args = '--low-memory --read-files-command zcat --out-chim-type Junctions --min-chim-segment 15 --x3'
+                // Ref: https://forums.developer.nvidia.com/t/problem-with-gpu/256825/6
+                // Parabricks’s rnafq2bam requires 24GB of memory.
+                // Using --low-memory for testing
+            }
+            process {
+                """
+                input[0] = Channel.of([
+                    [ id:'test', single_end:true ], // meta map
+                    [ file(params.modules_testdata_base_path + 'genomics/homo_sapiens/genome/test_starfusion_rnaseq_1.fastq.gz', checkIfExists: true) ]
+                ])
+                input[1] = Channel.of([
+                    [ id:'test' ], // meta map
+                    file(params.modules_testdata_base_path + 'genomics/homo_sapiens/genome/minigenome.gtf', checkIfExists: true)
+                ])
+				input[2] = BWA_INDEX.out.index
+                input[3] = PARABRICKS_STARGENOMEGENERATE.out.index
+                """
+            }
+        }
+
+        then {
+            assertAll(
+                { assert process.success },
+                { assert snapshot(
+                    bam(process.out.bam[0][1]).getReadsMD5(),
+                    file(process.out.bai[0][1]).name,
+                    file(process.out.junction[0][1]).name,
+                    process.out.versions,
+                    path(process.out.versions[0]).yaml
+                ).match() }
+            )
+        }
+    }
+
+    test("homo_sapiens_chimeric - stub") {
+
+        config "./nextflow.config"
+        options "-stub"
+
+        when {
+            params {
+                module_args = '--out-chim-type Junctions'
+            }
+            process {
+                """
+                input[0] = Channel.of([
+                    [ id:'test', single_end:true ], // meta map
+                    [ file(params.modules_testdata_base_path + 'genomics/homo_sapiens/genome/test_starfusion_rnaseq_1.fastq.gz', checkIfExists: true) ]
+                ])
+                input[1] = Channel.of([
+                    [ id:'test' ], // meta map
+                    file(params.modules_testdata_base_path + 'genomics/homo_sapiens/genome/minigenome.gtf', checkIfExists: true)
                 ])
 				input[2] = BWA_INDEX.out.index
                 input[3] = PARABRICKS_STARGENOMEGENERATE.out.index
@@ -113,8 +190,8 @@ nextflow_process {
                 { assert process.success },
                 { assert snapshot(
                     process.out,
-                    path(process.out.versions.get(0)).yaml,
-                    ).match() }
+                    path(process.out.versions[0]).yaml
+                ).match() }
             )
         }
     }