Merge pull request #108 from replikation/integration_Virsorter-2

Integration virsorter 2

README.md

@@ -276,6 +276,7 @@ Toolname/Gitlink | Reference |
 [VIBRANT](https://github.com/AnantharamanLab/VIBRANT)|[Automated recovery, annotation and curation of microbial viruses, and evaluation of virome function from genomic sequences](https://www.biorxiv.org/content/biorxiv/early/2019/11/26/855387.full.pdf)
 [VirNet](https://github.com/alyosama/virnet)|[Deep attention model for viral reads identification](https://ieeexplore.ieee.org/document/8639400)
 [Phigaro](https://github.com/bobeobibo/phigaro)| [Phigaro: high throughput prophage sequence annotation](https://www.biorxiv.org/content/10.1101/598243v1)
+[Virsorter2 beta](https://github.com/jiarong/VirSorter2)|
 
 
 

configs/container.config

@@ -21,4 +21,6 @@ process {
     withLabel: virnet           { container = 'multifractal/virnet-hack:0.1' }
     withLabel: virsorter        { container = 'multifractal/virsorter:0.1.2' }
     withLabel: phigaro          { container = 'multifractal/phigaro:0.5.2' }
+    withLabel: virsorter2       { container = 'multifractal/virsorter-2:0.1' }
+
 }

modules/parser/filter_virsorter2.nf

@@ -0,0 +1,14 @@
+process filter_virsorter2 {
+    label 'ubuntu'
+    input:
+        tuple val(name), path(results)
+    output:
+        tuple val(name), path("virsorter2_*.txt")
+    shell:
+        """
+        tail -n+2 *.tsv | awk '\$2>=${params.vs2_filter}' | awk '{ print \$1 }'  > virsorter2_\${PWD##*/}.txt
+        """
+}
+
+
+

modules/raw_data_collection/virsorter2_collect_data.nf

@@ -0,0 +1,14 @@
+process virsorter2_collect_data {
+    publishDir "${params.output}/${name}/raw_data", mode: 'copy', pattern: "virsorter2_results_${name}.tar.gz"
+    label 'ubuntu'
+    input:
+        tuple val(name), path(rawdir)
+    output:
+        tuple val(name), path("virsorter2_results_${name}.tar.gz")
+    script:
+        """
+        mkdir virsorter2
+        cp -r ${rawdir}/* virsorter2
+        tar -czf virsorter2_results_${name}.tar.gz virsorter2
+        """
+}

modules/tools/virsorter2.nf

@@ -0,0 +1,18 @@
+process virsorter2 {
+    label 'virsorter2'
+    errorStrategy 'ignore'
+    input:
+      tuple val(name), path(fasta)  
+    output:
+        tuple val(name), path("virsorter2_*.out/final-viral-score.tsv")
+        tuple val(name), path("virsorter2_*.out")
+    script:
+        """
+        virsorter run -w virsorter2_\${PWD##*/}.out -i ${fasta}  -j ${task.cpus}
+        """
+}
+
+
+//        tar cf virsorter2_results_\${PWD##*/}.out.tar virsorter2_\${PWD##*/}.out
+
+//

modules/upsetr.nf

@@ -32,7 +32,7 @@ process upsetr_plot {
 
       #nsets = 20, nintersects = 40
       upset(fromList(sets), sets = names(sets),
-          mainbar.y.label = "No. of shared viral contigs", sets.x.label = "Amount of identified \\nviral contigs", 
+          mainbar.y.label = "No. of shared viral contigs", sets.x.label = "Number of identified \\nviral contigs", 
           order.by = "freq", sets.bar.color = "#56B4E9", keep.order = F, 
           text.scale = 1.4, point.size = 2.6, line.size = 0.8, set_size.show = FALSE)
 

nextflow.config

@@ -32,6 +32,7 @@ params {
     mp = false
     vf = false
     vs = false
+    vs2 = false
     pp = false
     sm = false
     vb = false
@@ -49,7 +50,7 @@ params {
     vf_filter = '0.9'
     sm_filter = '0.5'
     vn_filter = '0.5'
-    // vs_filter = ''
+    vs2_filter = '0.9'
     // pp_filter = ''
     // vb_filter = ''
 

phage.nf

@@ -189,6 +189,9 @@ if (!params.setup && !workflow.profile.contains('test') && !workflow.profile.con
     include { virsorter_collect_data; virsorter_virome_collect_data } from './modules/raw_data_collection/virsorter_collect_data'
     include { phigaro } from './modules/tools/phigaro'
     include { phigaro_collect_data } from './modules/raw_data_collection/phigaro_collect_data'
+    include { virsorter2 } from './modules/tools/virsorter2'
+    include { filter_virsorter2 } from './modules/parser/filter_virsorter2'
+    include { virsorter2_collect_data} from './modules/raw_data_collection/virsorter2_collect_data'
 
 /************* 
 * DATABASES for Phage Identification
@@ -483,6 +486,22 @@ workflow virsorter_virome_wf {
             else { virsorter_virome_results = Channel.from( [ 'deactivated', 'deactivated'] ) }
     emit:   virsorter_virome_results
 } 
+//---------------------------------------------------------------------
+workflow virsorter2_wf {
+    take:   fasta           
+    main:   if (!params.vs2) {
+                        virsorter2(fasta)
+                        // filtering
+                        filter_virsorter2(virsorter2.out[0].groupTuple(remainder: true))
+                        // raw data collector
+                        virsorter2_collect_data(virsorter2.out[1].groupTuple(remainder: true))
+                        // result channel
+                        virsorter2_results = filter_virsorter2.out
+                        }
+            else { virsorter2_results = Channel.from( [ 'deactivated', 'deactivated'] ) }
+    emit:   virsorter2_results
+} 
+//------------------------------------------------------------------------
 
 workflow pprmeta_wf {
     take:   fasta
@@ -640,6 +659,7 @@ workflow identify_fasta_MSF {
 
         // gather results
             results =   virsorter_wf(fasta_validation_wf.out, virsorter_DB)
+                        .concat(virsorter2_wf(fasta_validation_wf.out))
                         .concat(virsorter_virome_wf(fasta_validation_wf.out, virsorter_DB))
                         .concat(marvel_wf(fasta_validation_wf.out))      
                         .concat(sourmash_wf(fasta_validation_wf.out, sourmash_DB))
@@ -828,7 +848,8 @@ def helpMSG() {
     --virome            deactivates virome-mode (vibrand and virsorter)
     --dv_filter         p-value cut-off [default: $params.dv_filter]
     --mp_filter         average nucleotide identity [default: $params.mp_filter]
-    --vf_filter         p-value cut-off [default: $params.vf_filter]
+    --vf_filter         score cut-off [default: $params.vf_filter]
+    --vs2_filter        dsDNAphage score cut-off [default: $params.vs2_filter]
     --sm_filter         Similarity score [default: $params.sm_filter]
     --vn_filter         Score [default: $params.vn_filter]