Merge branch 'clustersets' into dev

nf-core · Sep 22, 2024 · 553369f · 553369f
2 parents a419a6b + c5afe4c
commit 553369f
Show file tree

Hide file tree

Showing 17 changed files with 184 additions and 100 deletions.
diff --git a/conf/modules.config b/conf/modules.config
@@ -216,7 +216,7 @@ process {
         ext.args2 = '-f ID PRIMER ERROR PRSTART'
     }
 
-    withName: 'PRESTO_MASKPRIMERS_SCORE_SANSUMI_FWD' {
+    withName: PRESTO_MASKPRIMERS_SCORE_SANSUMI_FWD {
         publishDir = [
             path: { "${params.outdir}/presto/03-maskprimers/${meta.id}" },
             mode: params.publish_dir_mode,
@@ -225,7 +225,7 @@ process {
         ext.args2 = '-f ID PRIMER ERROR'
     }
 
-    withName: 'PRESTO_MASKPRIMERS_SCORE_SANSUMI_REV' {
+    withName: PRESTO_MASKPRIMERS_SCORE_SANSUMI_REV {
         publishDir = [
             path: { "${params.outdir}/presto/03-maskprimers/${meta.id}" },
             mode: params.publish_dir_mode,
@@ -234,6 +234,27 @@ process {
         ext.args2 = '-f ID PRIMER ERROR'
     }
 
+    withName: PRESTO_MASKPRIMERS_ALIGN_SANSUMI_FWD {
+        publishDir = [
+            path: { "${params.outdir}/presto/03-maskprimers/${meta.id}" },
+            mode: params.publish_dir_mode,
+            pattern: "*{txt,log,tab}"
+        ]
+        ext.args = '--pf FWD_PRIMER'
+        ext.args2 = '-f ID PRIMER ERROR'
+    }
+
+    withName: PRESTO_MASKPRIMERS_ALIGN_SANSUMI_REV {
+        publishDir = [
+            path: { "${params.outdir}/presto/03-maskprimers/${meta.id}" },
+            mode: params.publish_dir_mode,
+            pattern: "*{txt,log,tab}"
+        ]
+        ext.args = '--revpr --pf REV_PRIMER'
+        ext.args2 = '-f ID PRIMER ERROR'
+    }
+
+
     withName: PRESTO_PAIRSEQ {
         publishDir = [
             path: { "${params.outdir}/presto/03-pairseq/${meta.id}" },
@@ -261,12 +282,14 @@ process {
         ext.args = '--coord illumina'
     }
 
-    withName: PRESTO_CLUSTERSETS {
+    withName: PRESTO_CLUSTERSETS_UMI {
         publishDir = [
             path: { "${params.outdir}/presto/04-clustersets/${meta.id}" },
             mode: params.publish_dir_mode,
             pattern: "*{txt,log,tab}"
         ]
+        ext.args = '--exec vsearch --ident 0.9'
+        ext.args2 = '-f ID BARCODE SEQCOUNT CLUSTERS'
     }
 
     withName: PRESTO_PARSE_CLUSTER {
@@ -277,6 +300,15 @@ process {
         ]
     }
 
+    withName: PRESTO_PAIRSEQ_CLUSTERSETS {
+        publishDir = [
+            path: { "${params.outdir}/presto/05-parse-clusters/${meta.id}" },
+            mode: params.publish_dir_mode,
+            pattern: "*{txt,log,tab}"
+        ]
+        ext.args = '--coord illumina'
+    }
+
     withName: PRESTO_BUILDCONSENSUS_UMI {
         publishDir = [
             path: { "${params.outdir}/presto/06-build-consensus/${meta.id}" },
@@ -360,7 +392,7 @@ process {
             mode: params.publish_dir_mode,
             pattern: "*{txt,log,tab}"
         ]
-        ext.args = '--rc tail'
+        ext.args = '--rc tail --coord illumina'
         ext.args2 = '-f ID SEQCOUNT PRIMER PRCOUNT PRFREQ LENGTH OVERLAP ERROR PVALUE'
     }
 

diff --git a/conf/test_maskprimers_extract.config b/conf/test_maskprimers_extract.config
@@ -21,8 +21,6 @@ params {
 
     // Input data
     input = pipelines_testdata_base_path + 'testdata-bcr/Metadata_test_airr.tsv'
-    cprimers = pipelines_testdata_base_path + 'testdata-bcr/C_primers.fasta'
-    vprimers = pipelines_testdata_base_path + 'testdata-bcr/V_primers.fasta'
     reference_fasta = pipelines_testdata_base_path + 'database-cache/imgtdb_base.zip'
     reference_igblast = pipelines_testdata_base_path + 'database-cache/igblast_base.zip'
 

diff --git a/modules/local/presto/presto_assemblepairs.nf b/modules/local/presto/presto_assemblepairs.nf
@@ -14,7 +14,6 @@ process PRESTO_ASSEMBLEPAIRS {
     output:
     tuple val(meta), path("*_assemble-pass.fastq"), emit: reads
     path("*_command_log.txt"), emit: logs
-    path("*.log")
     path("*_table.tab")
     path "versions.yml" , emit: versions
 

diff --git a/modules/local/presto/presto_assemblepairs_sequential.nf b/modules/local/presto/presto_assemblepairs_sequential.nf
@@ -15,7 +15,6 @@ process PRESTO_ASSEMBLEPAIRS_SEQUENTIAL {
     output:
     tuple val(meta), path("*_assemble-pass.fastq"), emit: reads
     path("*_command_log.txt"), emit: logs
-    path("*.log")
     path("*_table.tab")
     path "versions.yml" , emit: versions
 

diff --git a/modules/local/presto/presto_buildconsensus.nf b/modules/local/presto/presto_buildconsensus.nf
@@ -14,8 +14,6 @@ process PRESTO_BUILDCONSENSUS {
     output:
     tuple val(meta), path("*_R1_consensus-pass.fastq"), path("*_R2_consensus-pass.fastq"), emit: reads
     path("*_command_log.txt"), emit: logs
-    path("*_R1.log")
-    path("*_R2.log")
     path("*.tab"), emit: log_tab
     path "versions.yml" , emit: versions
 

diff --git a/modules/local/presto/presto_clustersets.nf b/modules/local/presto/presto_clustersets.nf
@@ -14,15 +14,16 @@ process PRESTO_CLUSTERSETS {
     output:
     tuple val(meta), path("*_R1_cluster-pass.fastq"), path("*_R2_cluster-pass.fastq"), emit: reads
     path "*_command_log.txt", emit: logs
-    path "*.log"
     path "*.tab", emit: log_tab
     path("versions.yml"), emit: versions
 
     script:
+    def args = task.ext.args ?: ''
+    def args2 = task.ext.args2 ?: ''
     """
-    ClusterSets.py set --nproc ${task.cpus} -s $R1 --outname ${meta.id}_R1 --exec vsearch --log ${meta.id}_R1.log > ${meta.id}_command_log.txt
-    ClusterSets.py set --nproc ${task.cpus} -s $R2 --outname ${meta.id}_R2 --exec vsearch --log ${meta.id}_R2.log >> ${meta.id}_command_log.txt
-    ParseLog.py -l ${meta.id}_R1.log ${meta.id}_R2.log -f ID BARCODE SEQCOUNT CLUSTERS
+    ClusterSets.py set --nproc ${task.cpus} -s $R1 --outname ${meta.id}_R1 $args --log ${meta.id}_R1.log > ${meta.id}_command_log.txt
+    ClusterSets.py set --nproc ${task.cpus} -s $R2 --outname ${meta.id}_R2 $args --log ${meta.id}_R2.log >> ${meta.id}_command_log.txt
+    ParseLog.py -l ${meta.id}_R1.log ${meta.id}_R2.log $args2
 
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":

diff --git a/modules/local/presto/presto_collapseseq.nf b/modules/local/presto/presto_collapseseq.nf
@@ -14,7 +14,6 @@ process PRESTO_COLLAPSESEQ {
     output:
     tuple val(meta), path("*_collapse-unique.fastq") , emit: reads
     path("*_command_log.txt") , emit: logs
-    path("*.log")
     path("*_table.tab")
     path("versions.yml"), emit: versions
 

diff --git a/modules/local/presto/presto_filterseq.nf b/modules/local/presto/presto_filterseq.nf
@@ -15,8 +15,6 @@ process PRESTO_FILTERSEQ {
     tuple val(meta), path("*R1_quality-pass.fastq"), path("*R2_quality-pass.fastq") ,  emit: reads
     path "*_command_log_R?.txt" , emit: logs
     path "versions.yml" , emit: versions
-    path "*_R1.log"
-    path "*_R2.log"
     path "*.tab" , emit: log_tab
 
     script:

diff --git a/modules/local/presto/presto_filterseq_postassembly.nf b/modules/local/presto/presto_filterseq_postassembly.nf
@@ -15,7 +15,6 @@ process PRESTO_FILTERSEQ_POSTASSEMBLY {
     tuple val(meta), path("*quality-pass.fastq") ,  emit: reads
     path "*_command_log.txt" , emit: logs
     path "versions.yml" , emit: versions
-    path "*.log"
     path "*.tab" , emit: log_tab
 
     script:

diff --git a/modules/local/presto/presto_maskprimers.nf b/modules/local/presto/presto_maskprimers.nf
@@ -16,8 +16,6 @@ process PRESTO_MASKPRIMERS {
     output:
     tuple val(meta), path("*_R1_primers-pass.fastq"), path("*_R2_primers-pass.fastq") , emit: reads
     path "*_command_log_R?.txt", emit: logs
-    path "*_R1.log"
-    path "*_R2.log"
     path "*.tab", emit: log_tab
     path "versions.yml" , emit: versions
 

diff --git a/modules/local/presto/presto_maskprimers_align.nf b/modules/local/presto/presto_maskprimers_align.nf
@@ -10,15 +10,15 @@ process PRESTO_MASKPRIMERS_ALIGN {
 
     input:
     tuple val(meta), path(R1)
-    path(cprimers)
+    path(primers)
     val(max_len)
     val(max_error)
     val(mask_mode)
+    val(suffix)
 
     output:
-    tuple val(meta), path("*_R1_primers-pass.fastq") , emit: reads
-    path "*_command_log_R1.txt", emit: logs
-    path "*_R1.log"
+    tuple val(meta), path("*_primers-pass.fastq") , emit: reads
+    path "*.txt", emit: logs
     path "*.tab", emit: log_tab
     path "versions.yml" , emit: versions
 
@@ -28,14 +28,14 @@ process PRESTO_MASKPRIMERS_ALIGN {
     """
     MaskPrimers.py align --nproc ${task.cpus} \\
     -s $R1 \\
-    -p ${cprimers} \\
+    -p ${primers} \\
     --maxlen ${max_len} \\
     --maxerror ${max_error} \\
     --mode ${mask_mode} \\
     $args \\
-    --outname ${meta.id}_R1 \\
-    --log ${meta.id}_R1.log > ${meta.id}_command_log_R1.txt
-    ParseLog.py -l ${meta.id}_R1.log $args2
+    --outname ${meta.id}_${suffix} \\
+    --log ${meta.id}_R1.log > ${meta.id}_command_log_${suffix}.txt
+    ParseLog.py -l ${meta.id}_${suffix}.log $args2
 
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":

diff --git a/modules/local/presto/presto_maskprimers_extract.nf b/modules/local/presto/presto_maskprimers_extract.nf
@@ -19,7 +19,6 @@ process PRESTO_MASKPRIMERS_EXTRACT {
     output:
     tuple val(meta), path("*_primers-pass.fastq") , emit: reads
     path "*.txt", emit: logs
-    path "*.log"
     path "*.tab", emit: log_tab
     path "versions.yml" , emit: versions
 

diff --git a/modules/local/presto/presto_maskprimers_postassembly.nf b/modules/local/presto/presto_maskprimers_postassembly.nf
@@ -16,7 +16,6 @@ process PRESTO_MASKPRIMERS_POSTASSEMBLY {
     output:
     tuple val(meta), path("*REV_primers-pass.fastq") , emit: reads
     path "*command_log.txt", emit: logs
-    path "*.log"
     path "*.tab", emit: log_tab
     path "versions.yml" , emit: versions
 

diff --git a/modules/local/presto/presto_maskprimers_score.nf b/modules/local/presto/presto_maskprimers_score.nf
@@ -21,7 +21,6 @@ process PRESTO_MASKPRIMERS_SCORE {
     output:
     tuple val(meta), path("*_primers-pass.fastq"), emit: reads
     path "*.txt", emit: logs
-    path "*.log"
     path "*.tab", emit: log_tab
     path "versions.yml" , emit: versions