Nreport and chromosome

Author: replikation

configs/container.config

@@ -1,4 +1,5 @@
 process {
     // guppy container config is located in the guppy_gpu module
     withLabel: minimap2         { container = 'nanozoo/minimap2:2.26--d9ef6b6' }
+    withLabel:  plasflow        { container = 'quay.io/biocontainers/plasflow:1.1.0--py35_0' }
 }

configs/local.config

@@ -1,3 +1,4 @@
 process {
     withLabel: minimap2         { cpus = params.cores }
+    withLabel: plasflow         { cpus = params.cores }
 }

configs/nodes.config

@@ -1,3 +1,4 @@
 process {   
     withLabel: minimap2         { cpus = 24 ; memory = '32 GB' }
+    withLabel: plasflow         { cpus = 18 ; memory = '24 GB' }
 }

workflows/mask_regions.nf

@@ -1,4 +1,5 @@
 include { minimap2 } from './process/minimap2.nf'
+include { plasflow } from './process/plasflow.nf'
 
 workflow mask_regions_wf {
     take:   
@@ -13,11 +14,18 @@ workflow mask_regions_wf {
     no_match = combined_ch.ifEmpty{ log.info "\033[0;33mCould not match any reads to genomes, please read the help via --help\033[0m" }
 
     minimap2(combined_ch)
+    plasflow(minimap2.out.fasta)
 
     // report masked status
-    report_ch = minimap2.out.view { name, file, masked -> "Sample: $name has $masked masked bases" }
+    report_ch = plasflow.out.report.view { name, all, chr -> "$name: Total masked Bases $all with $chr on the chromosome only." }
 
-    emit: minimap2.out
+    report_write_ch = plasflow.out.report
+        .collectFile(seed: 'name,total masked bases,masked bases chromosome\n', 
+                    storeDir: params.output + "/") {
+                    row -> [ "masked_bases_summary.csv", row[0] + ',' + row[1] + ',' + row[2] + '\n']
+                    }
+
+    emit: minimap2.out.fasta
 
 
 }

workflows/process/minimap2.nf

@@ -4,18 +4,23 @@ process minimap2 {
     input:
         tuple val(name), path(fasta), path(reads)
   	output:
-    	tuple val(name), file("${name}.masked.fasta"), env(MASKREGIONS)
+    	tuple val(name), file("${name}.masked.fasta"), env(MASKREGIONS), emit: fasta
+        tuple val(name), file("${name}.masked.sorted.bam"), emit: bam
   	script:
     """
     minimap2 -t ${task.cpus} -o ${name}.sam -ax map-ont ${fasta} ${reads}
     samtools view -bS ${name}.sam | samtools sort - -@ ${task.cpus} -o ${name}.minimap.sorted.bam
 
     # consensus
     samtools consensus -@ ${task.cpus} -f fasta -X r10.4_sup ${name}.minimap.sorted.bam -o ${name}.masked.fasta
+    rm ${name}.minimap.sorted.bam
 
     # get Masked Bases
     MASKREGIONS=\$(grep -v ">" ${name}.masked.fasta | grep -o "N" | wc -l)
 
+    # rebam to masked file for visuals
+    minimap2 -t ${task.cpus} -o ${name}.masked.sam -ax map-ont ${name}.masked.fasta ${reads}
+    samtools view -bS ${name}.masked.sam | samtools sort - -@ ${task.cpus} -o ${name}.masked.sorted.bam
     """
     stub:
     """

workflows/process/plasflow.nf

@@ -0,0 +1,25 @@
+process plasflow {
+        label 'plasflow'
+        publishDir "${params.output}/${name}/", mode: 'copy'
+    input:
+        tuple val(name), path(fasta), val(allmasked) 
+    output:
+        tuple val(name), path("${name}_chromosomes.fasta"), emit: chromosomes  optional true
+        tuple val(name), path("${name}_plasmids.fasta"), emit: plasmids  optional true
+        tuple val(name), path("${name}_unclassified.fasta"), emit: unclassified  optional true
+        tuple val(name), val(allmasked), env(MASKREGIONS), emit: report
+    script:
+        """
+        PlasFlow.py --input ${fasta} --output ${name} --threshold 0.7
+
+        # remove empty file
+        find . -name "*.fasta" -type f -size 0 -print0 | xargs -0 echo rm
+
+        # get Masked Bases
+        MASKREGIONS=\$(grep -v ">" ${name}_chromosomes.fasta | grep -o "N" | wc -l)
+        """
+        stub:
+        """
+        touch  ${name}_chromosomes.fasta ${name}_plasmids.fasta ${name}_unclassified.fasta
+        """
+}