Polish STATS subworkflow

CHANGELOG.md

@@ -26,6 +26,7 @@ Initial release of nf-core/multiplesequencealign, created with the [nf-core](htt
 [#66](https://github.com/nf-core/multiplesequencealign/issues/66) - README: add new metromap and available tool list.
 [#54](https://github.com/nf-core/multiplesequencealign/issues/54) - Update modules versions from nf-core tools.
 [#80](https://github.com/nf-core/multiplesequencealign/pull/80) - Update modules versions from nf-core tools with nf-test.
+[#32](https://github.com/nf-core/multiplesequencealign/issues/32) - Update Stats workflow with nf-core modules for merging.
 
 ### `Dependencies`
 

assets/samplesheet.csv

@@ -1,3 +1,2 @@
 id,fasta,reference,structures,template
 seatoxin-ref,test-dataset/setoxin-ref.fa,test-dataset/setoxin.ref,https://raw.githubusercontent.com/nf-core/test-datasets/multiplesequencealign/testdata/structures/seatoxin-ref.tar.gz,
-toxin-ref,test-dataset/toxin-ref.fa,test-dataset/toxin.ref,

assets/toolsheet.csv

@@ -1,8 +1,4 @@
 tree,args_tree,aligner,args_aligner
 FAMSA,"-gt upgma -parttree",FAMSA,""
 "","",MAFFT,"--anysymbol --quiet --dpparttree"
-"","","KALIGN",""
-CLUSTALO,"--force",3DCOFFEE,"-method TMalign_pair"
-CLUSTALO,"--force",TCOFFEE,"-reg -reg_method famsa_msa -reg_nseq 100"
-"","",FAMSA,""
-FAMSA,"",TCOFFEE,"-reg -reg_method famsa_msa -reg_nseq 100"
+

bin/calc_seqstats.py

@@ -1,7 +1,6 @@
 #!/usr/bin/env python3
 from Bio import SeqIO
 import pandas as pd
-import time
 import sys
 
 fam_name = sys.argv[1]

conf/modules.config

@@ -18,6 +18,31 @@ process {
         saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
     ]
 
+    withName: TCOFFEE_SEQREFORMAT_SIM{
+        ext.args = "-output=sim_idscore"
+        publishDir = [
+            path: { "${params.outdir}/stats/${task.process.tokenize(':')[-1].toLowerCase()}" },
+            mode: params.publish_dir_mode,
+            saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
+        ]
+    }
+
+    withName: PARSE_SIM{
+        publishDir = [
+            path: { "${params.outdir}/stats/${task.process.tokenize(':')[-1].toLowerCase()}" },
+            mode: params.publish_dir_mode,
+            saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
+        ]
+    }
+
+    withName: "CALCULATE_SEQSTATS"{
+        publishDir = [
+            path: { "${params.outdir}/stats/${task.process.tokenize(':')[-1].toLowerCase()}" },
+            mode: params.publish_dir_mode,
+            saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
+        ]
+    }
+
     withName: "FAMSA_GUIDETREE"{
         ext.prefix = { "${meta.id}_${meta.tree}-args-${meta.argstree_clean}" }
         ext.args = { "${meta.args_tree}" == "null" ? '' : "${meta.args_tree}" }
@@ -38,6 +63,15 @@ process {
         ]
     }
 
+    withName: "CREATE_TCOFFEETEMPLATE"{
+        ext.prefix = { "${meta.id}" }
+        publishDir = [
+            path: { "${params.outdir}/alignment/templates" },
+            mode: params.publish_dir_mode,
+            saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
+        ]
+    }
+
     withName: ".*ALIGN"{
         ext.prefix = { "${meta.id}_${meta.tree}-args-${meta.argstree_clean}_${meta.aligner}-args-${meta.argsaligner_clean}" }
         ext.args = { "${meta.args_aligner}" == "null" ? '' : "${meta.args_aligner}" }
@@ -70,17 +104,29 @@ process {
         ]
     }
 
-    withName: "MERGE_STATS"{
+
+    withName: "CONCAT_SEQSTATS"{
+        ext.prefix = { "summary_seqstats" }
         publishDir = [
             path: { "${params.outdir}/stats/" },
             mode: params.publish_dir_mode,
             saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
         ]
     }
 
-    withName: "TCOFFEE_SEQREFORMAT_SIM|CALCULATE_SEQSTATS"{
+    withName: "CONCAT_SIMSTATS"{
+        ext.prefix = { "summary_simstats" }
         publishDir = [
-            path: { "${params.outdir}/stats/${task.process.tokenize(':')[-1].toLowerCase()}" },
+            path: { "${params.outdir}/stats/" },
+            mode: params.publish_dir_mode,
+            saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
+        ]
+    }
+
+    withName: "MERGE_STATS"{
+        ext.prefix = { "complete_summary_stats" }
+        publishDir = [
+            path: { "${params.outdir}/stats/" },
             mode: params.publish_dir_mode,
             saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
         ]

conf/test.config

@@ -20,7 +20,7 @@ params {
     max_time   = '6.h'
 
     // Stats
-    skip_stats = true
+    skip_stats = false
     skip_eval  = true
 
     // Input data

modules.json

@@ -15,6 +15,16 @@
                         "git_sha": "1f253ec05723293df7757af8769f8389b7a1884e",
                         "installed_by": ["modules"]
                     },
+                    "csvtk/concat": {
+                        "branch": "master",
+                        "git_sha": "911696ea0b62df80e900ef244d7867d177971f73",
+                        "installed_by": ["modules"]
+                    },
+                    "csvtk/join": {
+                        "branch": "master",
+                        "git_sha": "b2420a8bbd8af137380aa0a0c2e9a92456e5bb21",
+                        "installed_by": ["modules"]
+                    },
                     "custom/dumpsoftwareversions": {
                         "branch": "master",
                         "git_sha": "bba7e362e4afead70653f84d8700588ea28d0f9e",
@@ -65,6 +75,11 @@
                         "git_sha": "c83c78835ca6d7a55b3f200718d887cbc7149d37",
                         "installed_by": ["modules"]
                     },
+                    "tcoffee/seqreformat": {
+                        "branch": "master",
+                        "git_sha": "f759fd45ecabb40c761df1338a4bb3851171a7f7",
+                        "installed_by": ["modules"]
+                    },
                     "untar": {
                         "branch": "master",
                         "git_sha": "3f5420aa22e00bd030a2556dfdffc9e164ec0ec5",

modules/local/calculate_seqstats.nf

@@ -3,7 +3,10 @@ process CALCULATE_SEQSTATS {
     tag "$meta.id"
     label 'process_low'
 
-    container 'luisas/structural_regression:20'
+    conda "conda-forge::python=3.11.0   conda-forge::biopython=1.80 conda-forge::pandas=1.5.2"
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        'https://depot.galaxyproject.org/singularity/mulled-v2-27978155697a3671f3ef9aead4b5c823a02cc0b7:548df772fe13c0232a7eab1bc1deb98b495a05ab-0' :
+        'biocontainers/mulled-v2-27978155697a3671f3ef9aead4b5c823a02cc0b7:548df772fe13c0232a7eab1bc1deb98b495a05ab-0' }"
 
     input:
     tuple val(meta), path(fasta)

modules/local/parse_sim.nf

@@ -0,0 +1,30 @@
+process PARSE_SIM {
+    tag "$meta.id"
+    label 'process_low'
+
+    input:
+    tuple val(meta), path(infile)
+
+    output:
+    tuple val (meta), path("${prefix}.sim_tot"), emit: sim_tot
+
+    when:
+    task.ext.when == null || task.ext.when
+
+    script:
+    def args = task.ext.args ?: ''
+    prefix = task.ext.prefix ?: "${meta.id}"
+    """
+    echo "$prefix" > tmp
+    grep ^TOT $infile | cut -f4 >> tmp
+
+    echo "id,perc_sim" > ${prefix}.sim_tot
+    cat tmp | tr '\\n' ',' | awk 'gsub(/,\$/,x)' >>  ${prefix}.sim_tot
+    """
+
+    stub:
+    def prefix = task.ext.prefix ?: "${meta.id}"
+    """
+    touch ${prefix}.sim_tot
+    """
+}