Merge pull request #189 from nf-core/database-check

Adds (native groovy!) database check

conf/modules.config

@@ -12,14 +12,6 @@
 
 process {
 
-    withName: DATABASE_CHECK {
-        publishDir = [
-            path: { "${params.outdir}/pipeline_info" },
-            mode: params.publish_dir_mode,
-            saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
-        ]
-    }
-
     withName: FASTQC {
         ext.args = '--quiet'
         ext.prefix = { "${meta.id}_${meta.run_accession}_raw" }

modules/local/database_check.nf

@@ -1,29 +0,0 @@
-process DATABASE_CHECK {
-    tag "$databasesheet"
-    label 'process_single'
-
-    conda (params.enable_conda ? "conda-forge::python=3.8.3" : null)
-    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
-        'https://depot.galaxyproject.org/singularity/python:3.8.3' :
-        'quay.io/biocontainers/python:3.8.3' }"
-
-    input:
-    path databasesheet
-
-    output:
-    path '*.csv'       , emit: csv
-    path "versions.yml", emit: versions
-
-    when:
-    task.ext.when == null || task.ext.when
-
-    script: // This script is bundled with the pipeline, in nf-core/taxprofiler/bin/
-    """
-    cat $databasesheet >> database_sheet.valid.csv
-
-    cat <<-END_VERSIONS > versions.yml
-    "${task.process}":
-        python: \$(python --version | sed 's/Python //g')
-    END_VERSIONS
-    """
-}

subworkflows/local/db_check.nf

@@ -2,7 +2,6 @@
 // Check input samplesheet and get read channels
 //
 
-include { DATABASE_CHECK   } from '../../modules/local/database_check'
 include { UNTAR            } from '../../modules/nf-core/untar/main'
 
 workflow DB_CHECK {
@@ -10,15 +9,27 @@ workflow DB_CHECK {
     dbsheet // file: /path/to/dbsheet.csv
 
     main:
+    ch_versions = Channel.empty()
 
-    // TODO: make database sheet check
+    // special check to check _between_ rows, for which we must group rows together
-    // Checks:
+    // note: this will run in parallel to within-row validity, but we can assume this will run faster thus will fail first
-    // 1) no duplicates,
+    Channel.fromPath(dbsheet)
-    // 2) args do not have quotes, e.g. just `,,` and NOT `,"",`
+            .splitCsv ( header:true, sep:',' )
-    parsed_samplesheet = DATABASE_CHECK ( dbsheet )
+            .map {[it.tool, it.db_name] }
-        .csv
+            .groupTuple()
+            .map {
+                tool, db_name ->
+                    def unique_names = db_name.unique(false)
+                    if ( unique_names.size() < db_name.size() ) exit 1, "[nf-core/taxprofiler] ERROR: Each database for a tool must have a unique name, duplicated detected. Tool: ${tool}, Database name: ${unique_names}"
+            }
+
+    // normal checks for within-row validity, so can be moved to separate functions
+    parsed_samplesheet = Channel.fromPath(dbsheet)
         .splitCsv ( header:true, sep:',' )
-        .map { create_db_channels(it) }
+        .map {
+            validate_db_rows(it)
+            create_db_channels(it)
+        }
 
     ch_dbs_for_untar = parsed_samplesheet
         .branch {
@@ -29,12 +40,32 @@ workflow DB_CHECK {
     // TODO Filter to only run UNTAR on DBs of tools actually using?
     // TODO make optional whether to save
     UNTAR ( ch_dbs_for_untar.untar )
+    ch_versions = ch_versions.mix(UNTAR.out.versions.first())
 
     ch_final_dbs = ch_dbs_for_untar.skip.mix( UNTAR.out.untar )
 
     emit:
     dbs = ch_final_dbs                        // channel: [ val(meta), [ db ] ]
-    versions = DATABASE_CHECK.out.versions.mix(UNTAR.out.versions.first()) // channel: [ versions.yml ]
+    versions = ch_versions                   // channel: [ versions.yml ]
+}
+
+def validate_db_rows(LinkedHashMap row){
+
+        // check minimum number of columns
+        if (row.size() < 4) exit 1, "[nf-core/taxprofiler] ERROR: Invalid database input sheet - malformed row (e.g. missing column). See documentation for more information. Error in: ${row}"
+
+        // all columns there
+        def expected_headers = ['tool', 'db_name', 'db_params', 'db_path']
+        if ( !row.keySet().containsAll(expected_headers) ) exit 1, "[nf-core/taxprofiler] ERROR: Invalid database input sheet - malformed column names. Please check input TSV. Column names should be: ${expected_keys.join(", ")}"
+
+        // valid tools specified// TIFNISIH LIST
+        def expected_tools = [ "bracken", "centrifuge", "diamond", "kaiju",  "kraken2", "krakenuniq", "malt", "metaphlan3", "motus" ]
+        if ( !expected_tools.contains(row.tool) ) exit 1, "[nf-core/taxprofiler] ERROR: Invalid tool name. Please see documentation for all supported profilers. Error in: ${row}"
+
+        // detect quotes in params
+        if ( row.db_params.contains('"') ) exit 1, "[nf-core/taxprofiler] ERROR: Invalid database db_params entry. No quotes allowed. Error in: ${row}"
+        if ( row.db_params.contains("'") ) exit 1, "[nf-core/taxprofiler] ERROR: Invalid database db_params entry. No quotes allowed. Error in: ${row}"
+
 }
 
 def create_db_channels(LinkedHashMap row) {
@@ -45,9 +76,11 @@ def create_db_channels(LinkedHashMap row) {
 
     def array = []
     if (!file(row.db_path, type: 'dir').exists()) {
-        exit 1, "ERROR: Please check input samplesheet -> database could not be found!\n${row.db_path}"
+        exit 1, "ERROR: Please check input samplesheet -> database path could not be found!\n${row.db_path}"
     }
     array = [ meta, file(row.db_path) ]
 
     return array
 }
+
+

subworkflows/local/profiling.nf

@@ -41,14 +41,14 @@ workflow PROFILING {
             }
             .combine(databases)
             .branch {
-                malt:    it[2]['tool'] == 'malt'
-                kraken2: it[2]['tool'] == 'kraken2' || it[2]['tool'] == 'bracken' // to reuse the kraken module to produce the input data for bracken
-                metaphlan3: it[2]['tool'] == 'metaphlan3'
                 centrifuge: it[2]['tool'] == 'centrifuge'
-                kaiju: it[2]['tool'] == 'kaiju'
                 diamond: it[2]['tool'] == 'diamond'
-                motus: it[2]['tool'] == 'motus'
+                kaiju: it[2]['tool'] == 'kaiju'
+                kraken2: it[2]['tool'] == 'kraken2' || it[2]['tool'] == 'bracken' // to reuse the kraken module to produce the input data for bracken
                 krakenuniq: it[2]['tool'] == 'krakenuniq'
+                malt:    it[2]['tool'] == 'malt'
+                metaphlan3: it[2]['tool'] == 'metaphlan3'
+                motus: it[2]['tool'] == 'motus'
                 unknown: true
             }
 

workflows/taxprofiler.nf

@@ -25,7 +25,7 @@ if ( params.input ) {
     exit 1, "Input samplesheet, or PEP config and base directory not specified"
 }
 
-if (params.databases) { ch_databases = file(params.databases) } else { exit 1, 'Input database sheet not specified!' }
+if (params.databases) { ch_databases = file(params.databases, checkIfExists: true) } else { exit 1, 'Input database sheet not specified!' }
 
 if (params.shortread_qc_mergepairs && params.run_malt ) log.warn "[nf-core/taxprofiler] MALT does not accept uncollapsed paired-reads. Pairs will be profiled as separate files."
 if (params.shortread_qc_includeunmerged && !params.shortread_qc_mergepairs) exit 1, "ERROR: [nf-core/taxprofiler] cannot include unmerged reads when merging is not turned on. Please specify --shortread_qc_mergepairs"