fix: reduce number of required input files for damage profiler (#612)

* Reduce number of required input files for damage profiler * Remove rebugging * Add optional species list file. * Working pending updated test-dataset update * Add genome header to config
2024-12-22 11:08:17 +00:00 · 2021-10-20 10:02:30 +02:00 · 2021-10-20 10:02:30 +02:00 · 97fe899f79
commit 97fe899f79
parent 4e9e732b76
5 changed files with 144 additions and 35 deletions
--- a/modules/damageprofiler/main.nf
+++ b/modules/damageprofiler/main.nf
@ -22,25 +22,30 @@ process DAMAGEPROFILER {
    tuple val(meta), path(bam)
    path fasta
    path fai
+    path specieslist

    output:
    tuple val(meta), path("${prefix}"), emit: results
    path  "versions.yml"              , emit: versions

    script:
+    def software     = getSoftwareName(task.process)
    prefix           = options.suffix ? "${meta.id}${options.suffix}" : "${meta.id}"
+    def reference    = fasta ? "-r $fasta" : ""
+    def species_list = specieslist ? "-sf $specieslist" : ""

    """
    damageprofiler \\
    -i $bam \\
-        -r $fasta \\
    -o $prefix/ \\
-        $options.args
-
+    $options.args \\
+    $reference \\
+    $species_list

    cat <<-END_VERSIONS > versions.yml
    ${getProcessName(task.process)}:
        ${getSoftwareName(task.process)}: \$(damageprofiler -v | sed 's/^DamageProfiler v//')
    END_VERSIONS
    """
+
 }
--- a/modules/damageprofiler/meta.yml
+++ b/modules/damageprofiler/meta.yml
@ -32,12 +32,16 @@ input:
      pattern: "*.{bam,cram,sam}"
  - fasta:
      type: file
-      description: FASTA reference file
+      description: OPTIONAL FASTA reference file
      pattern: "*.{fasta,fna,fa}"
  - fai:
      type: file
-      description: FASTA index file from samtools faidx
+      description: OPTIONAL FASTA index file from samtools faidx
      pattern: "*.{fai}"
+  - specieslist:
+      type: file
+      description: OPTIONAL text file with list of target reference headers
+      pattern: "*.{txt}"

 output:
  - versions:
--- a/tests/config/test_data.config
+++ b/tests/config/test_data.config
@ -104,6 +104,7 @@ params {
                genome_gtf                                     = "${test_data_dir}/genomics/homo_sapiens/genome/genome.gtf"
                genome_sizes                                   = "${test_data_dir}/genomics/homo_sapiens/genome/genome.sizes"
                genome_bed                                     = "${test_data_dir}/genomics/homo_sapiens/genome/genome.bed"
+                genome_header                                  = "${test_data_dir}/genomics/homo_sapiens/genome/genome.header"
                genome_bed_gz                                  = "${test_data_dir}/genomics/homo_sapiens/genome/genome.bed.gz"
                genome_bed_gz_tbi                              = "${test_data_dir}/genomics/homo_sapiens/genome/genome.bed.gz.tbi"
                transcriptome_fasta                            = "${test_data_dir}/genomics/homo_sapiens/genome/transcriptome.fasta"
@ -123,6 +124,7 @@ params {
                test_paired_end_sorted_bam_bai                          = "${test_data_dir}/genomics/homo_sapiens/illumina/bam/test.paired_end.sorted.bam.bai"
                test_paired_end_markduplicates_sorted_bam               = "${test_data_dir}/genomics/homo_sapiens/illumina/bam/test.paired_end.markduplicates.sorted.bam"
                test_paired_end_markduplicates_sorted_bam_bai           = "${test_data_dir}/genomics/homo_sapiens/illumina/bam/test.paired_end.markduplicates.sorted.bam.bai"
+                test_paired_end_markduplicates_sorted_referencesn_txt   = "${test_data_dir}/genomics/homo_sapiens/illumina/bam/test.paired_end.markduplicates.sorted.referencesn.txt"
                test_paired_end_recalibrated_sorted_bam                 = "${test_data_dir}/genomics/homo_sapiens/illumina/bam/test.paired_end.recalibrated.sorted.bam"
                test_paired_end_recalibrated_sorted_bam_bai             = "${test_data_dir}/genomics/homo_sapiens/illumina/bam/test.paired_end.recalibrated.sorted.bam.bai"
                test_paired_end_umi_consensus_bam                       = "${test_data_dir}/genomics/homo_sapiens/illumina/bam/umi/test.paired_end.umi_consensus.bam"
--- a/tests/modules/damageprofiler/main.nf
+++ b/tests/modules/damageprofiler/main.nf
@ -7,9 +7,33 @@ include { DAMAGEPROFILER } from '../../../modules/damageprofiler/main.nf' addPar
 workflow test_damageprofiler {

    input        = [ [ id:'test', single_end:false ], // meta map
-              [ file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true) ] ]
+                   [ file(params.test_data['homo_sapiens']['illumina']['test_paired_end_markduplicates_sorted_bam'], checkIfExists: true) ] ]
+    fasta        = []
+    fai          = []
+    species_list = []
+
+
+    DAMAGEPROFILER ( input, fasta, fai, species_list )
+}
+
+workflow test_damageprofiler_reference {
+
+    input        = [ [ id:'test', single_end:false ], // meta map
+                   [ file(params.test_data['homo_sapiens']['illumina']['test_paired_end_markduplicates_sorted_bam'], checkIfExists: true) ] ]
    fasta        = file(params.test_data['homo_sapiens']['genome']['genome_fasta'], checkIfExists: true)
    fai          = file(params.test_data['homo_sapiens']['genome']['genome_fasta_fai'], checkIfExists: true)
+    species_list = []

-    DAMAGEPROFILER ( input, fasta, fai )
+    DAMAGEPROFILER ( input, fasta, fai, species_list )
+}
+
+workflow test_damageprofiler_specieslist {
+
+    input        = [ [ id:'test', single_end:false ], // meta map
+                   [ file(params.test_data['homo_sapiens']['illumina']['test_paired_end_markduplicates_sorted_bam'], checkIfExists: true) ] ]
+    fasta        = []
+    fai          = []
+    species_list = file(params.test_data['homo_sapiens']['genome']['genome_header'], checkIfExists: true)
+
+    DAMAGEPROFILER ( input, fasta, fai, species_list )
 }
--- a/tests/modules/damageprofiler/test.yml
+++ b/tests/modules/damageprofiler/test.yml
@ -4,13 +4,13 @@
    - damageprofiler
  files:
    - path: output/damageprofiler/test/3p_freq_misincorporations.txt
-      md5sum: da4cac90c78899a7cb6d72d415392b49
+      md5sum: de3b84d946a6b63cdcfadf82bf6854c0
    - path: output/damageprofiler/test/3pGtoA_freq.txt
-      md5sum: 8dab75d51a4b943b501d0995169c767f
+      md5sum: 61c903b1504ed7d7182570dfc75e4498
    - path: output/damageprofiler/test/5pCtoT_freq.txt
-      md5sum: fcc48ee5f72edff930d627c8bfdd8a5b
+      md5sum: 15a75b60ee519b61ce04a83fe3afe855
    - path: output/damageprofiler/test/5p_freq_misincorporations.txt
-      md5sum: 54665474f5ef17dcc268567e5eaa7d86
+      md5sum: 3b3240d6c1a3491e461b39199a9fcfe3
    - path: output/damageprofiler/test/DamagePlot_five_prime.svg
    - path: output/damageprofiler/test/DamagePlot.pdf
    - path: output/damageprofiler/test/DamagePlot_three_prime.svg
@ -18,19 +18,93 @@
      contains:
        - "FINISHED SUCCESSFULLY"
    - path: output/damageprofiler/test/dmgprof.json
-      md5sum: 98499024c7e937896e481f2d3cfbdd3e
+      md5sum: 2e54e712d2ae9e32c4c298e5fd8f60fe
    - path: output/damageprofiler/test/DNA_comp_genome.txt
-      md5sum: f91e70760d91a1193a27e360aaddf2fd
+      md5sum: fea48af1ecf491b439d36d4a919473df
    - path: output/damageprofiler/test/DNA_composition_sample.txt
-      md5sum: 1257eb3eb42484647bfba2151f9ef04f
+      md5sum: 9e17a0b1e5ad4eb13201cd24ad8507dd
    - path: output/damageprofiler/test/edit_distance.pdf
    - path: output/damageprofiler/test/edit_distance.svg
    - path: output/damageprofiler/test/editDistance.txt
-      md5sum: af2d2f4a99058ec56eae88ec27779e38
+      md5sum: 04d14b449a5afa8b5dbff0dfa762356b
    - path: output/damageprofiler/test/Length_plot_combined_data.svg
    - path: output/damageprofiler/test/Length_plot_forward_reverse_separated.svg
    - path: output/damageprofiler/test/Length_plot.pdf
    - path: output/damageprofiler/test/lgdistribution.txt
-      md5sum: c5d029bf3a92b613310ee23f47d94981
+      md5sum: df2e19195185ea9ee05e8e84b2948f36
    - path: output/damageprofiler/test/misincorporation.txt
-      md5sum: 3aa6dd749010a492d92a815a83c196a8
+      md5sum: bec0c5fc2fa9c82b04949e2d8b6e979c
+
+- name: damageprofiler_reference
+  command: nextflow run ./tests/modules/damageprofiler -entry test_damageprofiler_reference -c tests/config/nextflow.config -dump-channels
+  tags:
+    - damageprofiler
+  files:
+    - path: output/damageprofiler/test/3p_freq_misincorporations.txt
+      md5sum: de3b84d946a6b63cdcfadf82bf6854c0
+    - path: output/damageprofiler/test/3pGtoA_freq.txt
+      md5sum: 61c903b1504ed7d7182570dfc75e4498
+    - path: output/damageprofiler/test/5pCtoT_freq.txt
+      md5sum: 15a75b60ee519b61ce04a83fe3afe855
+    - path: output/damageprofiler/test/5p_freq_misincorporations.txt
+      md5sum: 3b3240d6c1a3491e461b39199a9fcfe3
+    - path: output/damageprofiler/test/DamagePlot_five_prime.svg
+    - path: output/damageprofiler/test/DamagePlot.pdf
+    - path: output/damageprofiler/test/DamagePlot_three_prime.svg
+    - path: output/damageprofiler/test/DamageProfiler.log
+      contains:
+        - "FINISHED SUCCESSFULLY"
+    - path: output/damageprofiler/test/dmgprof.json
+      md5sum: 2e54e712d2ae9e32c4c298e5fd8f60fe
+    - path: output/damageprofiler/test/DNA_comp_genome.txt
+      md5sum: fea48af1ecf491b439d36d4a919473df
+    - path: output/damageprofiler/test/DNA_composition_sample.txt
+      md5sum: 9e17a0b1e5ad4eb13201cd24ad8507dd
+    - path: output/damageprofiler/test/edit_distance.pdf
+    - path: output/damageprofiler/test/edit_distance.svg
+    - path: output/damageprofiler/test/editDistance.txt
+      md5sum: 04d14b449a5afa8b5dbff0dfa762356b
+    - path: output/damageprofiler/test/Length_plot_combined_data.svg
+    - path: output/damageprofiler/test/Length_plot_forward_reverse_separated.svg
+    - path: output/damageprofiler/test/Length_plot.pdf
+    - path: output/damageprofiler/test/lgdistribution.txt
+      md5sum: df2e19195185ea9ee05e8e84b2948f36
+    - path: output/damageprofiler/test/misincorporation.txt
+      md5sum: bec0c5fc2fa9c82b04949e2d8b6e979c
+
+- name: damageprofiler_specieslist
+  command: nextflow run ./tests/modules/damageprofiler -entry test_damageprofiler_specieslist -c tests/config/nextflow.config -dump-channels
+  tags:
+    - damageprofiler
+  files:
+    - path: output/damageprofiler/test/chr22/3p_freq_misincorporations.txt
+      md5sum: de3b84d946a6b63cdcfadf82bf6854c0
+    - path: output/damageprofiler/test/chr22/3pGtoA_freq.txt
+      md5sum: 61c903b1504ed7d7182570dfc75e4498
+    - path: output/damageprofiler/test/chr22/5pCtoT_freq.txt
+      md5sum: 15a75b60ee519b61ce04a83fe3afe855
+    - path: output/damageprofiler/test/chr22/5p_freq_misincorporations.txt
+      md5sum: 3b3240d6c1a3491e461b39199a9fcfe3
+    - path: output/damageprofiler/test/chr22/DamagePlot_five_prime.svg
+    - path: output/damageprofiler/test/chr22/DamagePlot.pdf
+    - path: output/damageprofiler/test/chr22/DamagePlot_three_prime.svg
+    - path: output/damageprofiler/test/DamageProfiler.log
+      contains:
+        - "FINISHED SUCCESSFULLY"
+    - path: output/damageprofiler/test/chr22/dmgprof.json
+      md5sum: 2e54e712d2ae9e32c4c298e5fd8f60fe
+    - path: output/damageprofiler/test/chr22/DNA_comp_genome.txt
+      md5sum: fea48af1ecf491b439d36d4a919473df
+    - path: output/damageprofiler/test/chr22/DNA_composition_sample.txt
+      md5sum: 9e17a0b1e5ad4eb13201cd24ad8507dd
+    - path: output/damageprofiler/test/chr22/edit_distance.pdf
+    - path: output/damageprofiler/test/chr22/edit_distance.svg
+    - path: output/damageprofiler/test/chr22/editDistance.txt
+      md5sum: 04d14b449a5afa8b5dbff0dfa762356b
+    - path: output/damageprofiler/test/chr22/Length_plot_combined_data.svg
+    - path: output/damageprofiler/test/chr22/Length_plot_forward_reverse_separated.svg
+    - path: output/damageprofiler/test/chr22/Length_plot.pdf
+    - path: output/damageprofiler/test/chr22/lgdistribution.txt
+      md5sum: df2e19195185ea9ee05e8e84b2948f36
+    - path: output/damageprofiler/test/chr22/misincorporation.txt
+      md5sum: bec0c5fc2fa9c82b04949e2d8b6e979c