Merge pull request #11 from TRON-Bioinformatics/develop

Merge Gianna's work into master + change CI environment requirements and conda/mamba setup
TRON-Bioinformatics · May 23, 2024 · b47efca · b47efca
2 parents 24e23fb + a9e8dd6
commit b47efca
Show file tree

Hide file tree

Showing 22 changed files with 98 additions and 85 deletions.
diff --git a/.github/workflows/automated_tests.yml b/.github/workflows/automated_tests.yml
@@ -12,7 +12,7 @@ jobs:
       with:
         distribution: 'zulu' # See 'Supported distributions' for available options
         java-version: '11'
-    - uses: conda-incubator/setup-miniconda@v2
+    - uses: mamba-org/setup-micromamba@v1
     - name: Install dependencies
       run: |
         apt-get update && apt-get --assume-yes install wget make procps software-properties-common

diff --git a/main.nf b/main.nf
@@ -78,25 +78,25 @@ workflow {
 
         // alignment
         if (params.algorithm == "aln" && !params.inception) {
-            BWA_ALN(trimmed_fastqs.map {name, fq1, fq2 -> tuple(name, fq1)})
-            BWA_ALN_2(trimmed_fastqs.map {name, fq1, fq2 -> tuple(name, fq2)})
-            BWA_SAMPE(BWA_ALN.out.alignment_output.join(BWA_ALN_2.out.alignment_output))
+            BWA_ALN(trimmed_fastqs.map {name, fq1, fq2 -> tuple(name, fq1)},params.reference)
+            BWA_ALN_2(trimmed_fastqs.map {name, fq1, fq2 -> tuple(name, fq2)},params.reference)
+            BWA_SAMPE(BWA_ALN.out.alignment_output.join(BWA_ALN_2.out.alignment_output),params.reference)
             output_bams = BWA_SAMPE.out.bams
         }
         else if (params.algorithm == "aln" && params.inception) {
-            BWA_ALN_INCEPTION(trimmed_fastqs)
+            BWA_ALN_INCEPTION(trimmed_fastqs,params.reference)
             output_bams = BWA_ALN_INCEPTION.out.bams
         }
         else if (params.algorithm == "mem") {
-            BWA_MEM(trimmed_fastqs)
+            BWA_MEM(trimmed_fastqs,params.reference)
             output_bams = BWA_MEM.out.bams
         }
         else if (params.algorithm == "mem2") {
-            BWA_MEM_2(trimmed_fastqs)
+            BWA_MEM_2(trimmed_fastqs,params.reference)
             output_bams = BWA_MEM_2.out.bams
         }
         else if (params.algorithm == "star") {
-            STAR(trimmed_fastqs)
+            STAR(trimmed_fastqs,params.reference)
             output_bams = STAR.out.bams
         }
         else {
@@ -112,20 +112,20 @@ workflow {
             trimmed_fastqs = FASTP_SINGLE.out.trimmed_fastqs
         }
         if (params.algorithm == "aln"  && !params.inception) {
-            BWA_ALN(trimmed_fastqs)
-            BWA_SAMSE(BWA_ALN.out.alignment_output)
+            BWA_ALN(trimmed_fastqs,params.reference)
+            BWA_SAMSE(BWA_ALN.out.alignment_output,params.reference)
             output_bams = BWA_SAMSE.out.bams
         }
         else if (params.algorithm == "mem") {
-            BWA_MEM_SE(trimmed_fastqs)
+            BWA_MEM_SE(trimmed_fastqs,params.reference)
             output_bams = BWA_MEM_SE.out.bams
         }
         else if (params.algorithm == "mem2") {
-            BWA_MEM_2_SE(trimmed_fastqs)
+            BWA_MEM_2_SE(trimmed_fastqs,params.reference)
             output_bams = BWA_MEM_2_SE.out.bams
         }
         else if (params.algorithm == "star") {
-            STAR_SE(trimmed_fastqs)
+            STAR_SE(trimmed_fastqs,params.reference)
             output_bams = STAR_SE.out.bams
         }
         else {

diff --git a/modules/01_fastp.nf b/modules/01_fastp.nf
@@ -16,7 +16,7 @@ process FASTP_PAIRED {
             file("${fastq2.baseName}.trimmed.fq.gz"), emit: trimmed_fastqs
         file("${name}.fastp_stats.json")
         file("${name}.fastp_stats.html")
-        file("software_versions.${task.process}.txt")
+        file("software_versions.FASTP_PAIRED.txt")
 
     """
     # --input_files needs to be forced, otherwise it is inherited from profile in tests
@@ -29,8 +29,8 @@ process FASTP_PAIRED {
     --html ${name}.fastp_stats.html \
     --thread ${params.cpus}
 
-    echo ${params.manifest} >> software_versions.${task.process}.txt
-    fastp --version 2>> software_versions.${task.process}.txt
+    echo ${params.manifest} >> software_versions.FASTP_PAIRED.txt
+    fastp --version 2>> software_versions.FASTP_PAIRED.txt
     """
 }
 
@@ -50,7 +50,7 @@ process FASTP_SINGLE {
         tuple val(name), file("${fastq1.baseName}.trimmed.fq.gz"), emit: trimmed_fastqs
         file("${name}.fastp_stats.json")
         file("${name}.fastp_stats.html")
-        file("software_versions.${task.process}.txt")
+        file("software_versions.FASTP_SINGLE.txt")
 
     """
     # --input_files needs to be forced, otherwise it is inherited from profile in tests
@@ -61,7 +61,7 @@ process FASTP_SINGLE {
     --html ${name}.fastp_stats.html \
     --thread ${params.cpus}
 
-    echo ${params.manifest} >> software_versions.${task.process}.txt
-    fastp --version 2>> software_versions.${task.process}.txt
+    echo ${params.manifest} >> software_versions.FASTP_SINGLE.txt
+    fastp --version 2>> software_versions.FASTP_SINGLE.txt
     """
 }
diff --git a/modules/02_bwa_aln.nf b/modules/02_bwa_aln.nf
@@ -9,16 +9,17 @@ process BWA_ALN {
 
     input:
         tuple val(name), file(fastq)
+        val(reference)
 
     output:
         tuple val("${name}"), file("${fastq}"), file("${fastq.baseName}.sai"), emit: alignment_output
-        file("software_versions.${task.process}.txt")
+        file("software_versions.BWA_ALN.txt")
 
     """
-    bwa aln -t ${task.cpus} ${params.reference} ${fastq} > ${fastq.baseName}.sai
+    bwa aln -t ${task.cpus} ${reference} ${fastq} > ${fastq.baseName}.sai
 
-    echo ${params.manifest} >> software_versions.${task.process}.txt
-    echo "bwa=0.7.17"  >> software_versions.${task.process}.txt
+    echo ${params.manifest} >> software_versions.BWA_ALN.txt
+    echo "bwa=0.7.17"  >> software_versions.BWA_ALN.txt
     """
 }
 
@@ -34,17 +35,18 @@ process BWA_SAMPE {
     input:
       // joins both channels by key using the first element in the tuple, the name
       tuple val(name), file(fastq1), file(sai1), file(fastq2), file(sai2)
+      val(reference)
 
     output:
       tuple val("${name}"), file("${name}.bam"), emit: bams
-      file("software_versions.${task.process}.txt")
+      file("software_versions.BWA_SAMPE.txt")
 
     """
-    bwa sampe ${params.reference} ${sai1} ${sai2} ${fastq1} ${fastq2} | samtools view -uS - | samtools sort - > ${name}.bam
+    bwa sampe ${reference} ${sai1} ${sai2} ${fastq1} ${fastq2} | samtools view -uS - | samtools sort - > ${name}.bam
 
-    echo ${params.manifest} >> software_versions.${task.process}.txt
-    echo "bwa=0.7.17"  >> software_versions.${task.process}.txt
-    samtools --version >> software_versions.${task.process}.txt
+    echo ${params.manifest} >> software_versions.BWA_SAMPE.txt
+    echo "bwa=0.7.17"  >> software_versions.BWA_SAMPE.txt
+    samtools --version >> software_versions.BWA_SAMPE.txt
     """
 }
 
@@ -60,16 +62,17 @@ process BWA_SAMSE {
     input:
       // joins both channels by key using the first element in the tuple, the name
       tuple val(name), file(fastq), file(sai)
+      val(reference)
 
     output:
       tuple val("${name}"), file("${name}.bam"), emit: bams
-      file("software_versions.${task.process}.txt")
+      file("software_versions.BWA_SAMSE.txt")
 
     """
-    bwa samse ${params.reference} ${sai} ${fastq} | samtools view -uS - | samtools sort - > ${name}.bam
+    bwa samse ${reference} ${sai} ${fastq} | samtools view -uS - | samtools sort - > ${name}.bam
 
-    echo ${params.manifest} >> software_versions.${task.process}.txt
-    echo "bwa=0.7.17"  >> software_versions.${task.process}.txt
+    echo ${params.manifest} >> software_versions.BWA_SAMSE.txt
+    echo "bwa=0.7.17"  >> software_versions.BWA_SAMSE.txt
     """
 }
 
@@ -86,18 +89,19 @@ process BWA_ALN_INCEPTION {
     input:
       // joins both channels by key using the first element in the tuple, the name
       tuple val(name), file(fastq1), file(fastq2)
+      val(reference)
 
     output:
       tuple val("${name}"), file("${name}.bam"), emit: bams
-      file("software_versions.${task.process}.txt")
+      file("software_versions.BWA_ALN_INCEPTION.txt")
 
     """
-    bwa sampe ${params.reference} <( bwa aln -t ${params.cpus} ${params.reference} ${fastq1} ) \
-    <( bwa aln -t ${params.cpus} ${params.reference} ${fastq2} ) ${fastq1} ${fastq2} \
+    bwa sampe ${reference} <( bwa aln -t ${params.cpus} ${reference} ${fastq1} ) \
+    <( bwa aln -t ${params.cpus} ${reference} ${fastq2} ) ${fastq1} ${fastq2} \
     | samtools view -uS - | samtools sort - > ${name}.bam
 
-    echo ${params.manifest} >> software_versions.${task.process}.txt
-    echo "bwa=0.7.17"  >> software_versions.${task.process}.txt
-    samtools --version >> software_versions.${task.process}.txt
+    echo ${params.manifest} >> software_versions.BWA_ALN_INCEPTION.txt
+    echo "bwa=0.7.17"  >> software_versions.BWA_ALN_INCEPTION.txt
+    samtools --version >> software_versions.BWA_ALN_INCEPTION.txt
     """
 }
diff --git a/modules/02_bwa_mem.nf b/modules/02_bwa_mem.nf
@@ -10,17 +10,18 @@ process BWA_MEM {
 
     input:
       tuple val(name), file(fastq1), file(fastq2)
+      val(reference)
 
     output:
       tuple val("${name}"), file("${name}.bam"), emit: bams
-      file("software_versions.${task.process}.txt")
+      file("software_versions.BWA_MEM.txt")
 
     """
-    bwa mem ${params.additional_args} -t ${task.cpus} ${params.reference} ${fastq1} ${fastq2} | samtools view -uS - | samtools sort - > ${name}.bam
+    bwa mem ${params.additional_args} -t ${task.cpus} ${reference} ${fastq1} ${fastq2} | samtools view -uS - | samtools sort - > ${name}.bam
 
-    echo ${params.manifest} >> software_versions.${task.process}.txt
-    echo "bwa=0.7.17"  >> software_versions.${task.process}.txt
-    samtools --version >> software_versions.${task.process}.txt
+    echo ${params.manifest} >> software_versions.BWA_MEM.txt
+    echo "bwa=0.7.17"  >> software_versions.BWA_MEM.txt
+    samtools --version >> software_versions.BWA_MEM.txt
     """
 }
 
@@ -36,16 +37,17 @@ process BWA_MEM_SE {
     input:
       // joins both channels by key using the first element in the tuple, the name
       tuple val(name), file(fastq)
+      val(reference)
 
     output:
       tuple val("${name}"), file("${name}.bam"), emit: bams
-      file("software_versions.${task.process}.txt")
+      file("software_versions.BWA_MEM_SE.txt")
 
     """
-    bwa mem ${params.additional_args} -t ${task.cpus} ${params.reference} ${fastq} | samtools view -uS - | samtools sort - > ${name}.bam
+    bwa mem ${params.additional_args} -t ${task.cpus} ${reference} ${fastq} | samtools view -uS - | samtools sort - > ${name}.bam
 
-    echo ${params.manifest} >> software_versions.${task.process}.txt
-    echo "bwa=0.7.17"  >> software_versions.${task.process}.txt
-    samtools --version >> software_versions.${task.process}.txt
+    echo ${params.manifest} >> software_versions.BWA_MEM_SE.txt
+    echo "bwa=0.7.17"  >> software_versions.BWA_MEM_SE.txt
+    samtools --version >> software_versions.BWA_MEM_SE.txt
     """
 }
diff --git a/modules/02_bwa_mem_2.nf b/modules/02_bwa_mem_2.nf
@@ -10,17 +10,18 @@ process BWA_MEM_2 {
 
     input:
       tuple val(name), file(fastq1), file(fastq2)
+      val(reference)
 
     output:
       tuple val("${name}"), file("${name}.bam"), emit: bams
-      file("software_versions.${task.process}.txt")
+      file("software_versions.BWA_MEM_2.txt")
 
     """
-    bwa-mem2 mem ${params.additional_args} -t ${task.cpus} ${params.reference} ${fastq1} ${fastq2} | samtools view -uS - | samtools sort - > ${name}.bam
+    bwa-mem2 mem ${params.additional_args} -t ${task.cpus} ${reference} ${fastq1} ${fastq2} | samtools view -uS - | samtools sort - > ${name}.bam
 
-    echo ${params.manifest} >> software_versions.${task.process}.txt
-    bwa-mem2 version  >> software_versions.${task.process}.txt
-    samtools --version >> software_versions.${task.process}.txt
+    echo ${params.manifest} >> software_versions.BWA_MEM_2.txt
+    bwa-mem2 version  >> software_versions.BWA_MEM_2.txt
+    samtools --version >> software_versions.BWA_MEM_2.txt
     """
 }
 
@@ -36,16 +37,17 @@ process BWA_MEM_2_SE {
     input:
       // joins both channels by key using the first element in the tuple, the name
       tuple val(name), file(fastq)
+      val(reference)
 
     output:
       tuple val("${name}"), file("${name}.bam"), emit: bams
-      file("software_versions.${task.process}.txt")
+      file("software_versions.BWA_MEM_2_SE.txt")
 
     """
-    bwa-mem2 mem ${params.additional_args} -t ${task.cpus} ${params.reference} ${fastq} | samtools view -uS - | samtools sort - > ${name}.bam
+    bwa-mem2 mem ${params.additional_args} -t ${task.cpus} ${reference} ${fastq} | samtools view -uS - | samtools sort - > ${name}.bam
 
-    echo ${params.manifest} >> software_versions.${task.process}.txt
-    bwa-mem2 version  >> software_versions.${task.process}.txt
-    samtools --version >> software_versions.${task.process}.txt
+    echo ${params.manifest} >> software_versions.BWA_MEM_2_SE.txt
+    bwa-mem2 version  >> software_versions.BWA_MEM_2_SE.txt
+    samtools --version >> software_versions.BWA_MEM_2_SE.txt
     """
 }
diff --git a/modules/02_star.nf b/modules/02_star.nf
@@ -10,15 +10,16 @@ process STAR {
 
     input:
       tuple val(name), file(fastq1), file(fastq2)
+      val(reference)
 
     output:
       tuple val("${name}"), file("${name}.bam"), emit: bams
-      file("software_versions.${task.process}.txt")
+      file("software_versions.STAR.txt")
 
     script:
     two_pass_mode_param = params.star_two_pass_mode ? "--twopassMode Basic" : ""
     """
-    STAR --genomeDir ${params.reference} ${two_pass_mode_param} ${params.additional_args} \
+    STAR --genomeDir ${reference} ${two_pass_mode_param} ${params.additional_args} \
     --readFilesCommand "gzip -d -c -f" \
     --readFilesIn ${fastq1} ${fastq2} \
     --outSAMmode Full \
@@ -32,8 +33,8 @@ process STAR {
 
     mv ${name}.Aligned.sortedByCoord.out.bam ${name}.bam
 
-    echo ${params.manifest} >> software_versions.${task.process}.txt
-    STAR --version >> software_versions.${task.process}.txt
+    echo ${params.manifest} >> software_versions.STAR.txt
+    STAR --version >> software_versions.STAR.txt
     """
 }
 
@@ -48,15 +49,16 @@ process STAR_SE {
 
     input:
       tuple val(name), file(fastq)
+      val(reference)
 
     output:
       tuple val("${name}"), file("${name}.bam"), emit: bams
-      file("software_versions.${task.process}.txt")
+      file("software_versions.STAR_SE.txt")
 
     script:
     two_pass_mode_param = params.star_two_pass_mode ? "--twopassMode Basic" : ""
     """
-    STAR --genomeDir ${params.reference} ${two_pass_mode_param} ${params.additional_args} \
+    STAR --genomeDir ${reference} ${two_pass_mode_param} ${params.additional_args} \
     --readFilesCommand "gzip -d -c -f" \
     --readFilesIn ${fastq} \
     --outSAMmode Full \
@@ -70,7 +72,7 @@ process STAR_SE {
 
     mv ${name}.Aligned.sortedByCoord.out.bam ${name}.bam
 
-    echo ${params.manifest} >> software_versions.${task.process}.txt
-    STAR --version >> software_versions.${task.process}.txt
+    echo ${params.manifest} >> software_versions.STAR_SE.txt
+    STAR --version >> software_versions.STAR_SE.txt
     """
 }
diff --git a/modules/03_index.nf b/modules/03_index.nf
@@ -13,12 +13,12 @@ process INDEX_BAM {
 
     output:
       file("${name}.bam.bai")
-      file("software_versions.${task.process}.txt")
+      file("software_versions.INDEX_BAM.txt")
 
     """
     samtools index -@ ${task.cpus} ${bam}
 
-    echo ${params.manifest} >> software_versions.${task.process}.txt
-    samtools --version >> software_versions.${task.process}.txt
+    echo ${params.manifest} >> software_versions.INDEX_BAM.txt
+    samtools --version >> software_versions.INDEX_BAM.txt
     """
 }
diff --git a/nextflow.config b/nextflow.config
@@ -22,18 +22,21 @@ params.additional_args = ""
 profiles {
   conda {
     params.enable_conda = true
+    conda.enabled = true
   }
   debug { process.beforeScript = 'echo $HOSTNAME' }
-  test {
-    params.input_files = "test_data/test_input_paired.txt"
-    params.reference = "$baseDir/test_data/ucsc.hg19.minimal.fasta"
+  ci {
     params.cpus = 1
     params.memory = "3g"
     timeline.enabled = false
     report.enabled = false
     trace.enabled = false
     dag.enabled = false
   }
+  test {
+    params.input_files = "test_data/test_input_paired.txt"
+    params.reference = "$baseDir/test_data/ucsc.hg19.minimal.fasta"
+  }
 }
 
 // Export this variable to prevent local Python libraries from conflicting with those in the container