add sidle reference taxonomy entries & custom input

nf-core · Feb 9, 2024 · afbb18f · afbb18f
1 parent aa1ff5a
commit afbb18f
Show file tree

Hide file tree

Showing 7 changed files with 166 additions and 37 deletions.
diff --git a/bin/taxref_reformat_sidle.sh b/bin/taxref_reformat_sidle.sh
@@ -0,0 +1,22 @@
+#!/bin/sh
+
+# Untar any tar file in the working directory
+tar xzf database.tar.gz
+
+# Greengenes 13_8
+if [ -d "gg_13_8_otus" ]; then
+    mv gg_13_8_otus/rep_set/99_otus.fasta gg_13_8_otus_rep_set_99_otus.seq.fasta
+    mv gg_13_8_otus/rep_set_aligned/99_otus.fasta gg_13_8_otus_rep_set_aligned_99_otus.alnseq.fasta
+    mv gg_13_8_otus/taxonomy/99_otu_taxonomy.txt gg_13_8_otus_taxonomy_99_otu_taxonomy.tax.txt
+elif [ -d "gg_13_8_otus" ]; then
+    mv SILVA_128_QIIME_release/rep_set/rep_set_all/99/99_otus.fasta SILVA_128_QIIME_release_rep_set_all_99_otus.seq.fasta
+    gunzip -c /SILVA_128_QIIME_release/rep_set_aligned/99/99_otus_aligned.fasta.gz > SILVA_128_QIIME_release_rep_set_aligned_99_otus_aligned.alnseq.fasta
+    mv SILVA_128_QIIME_release/taxonomy/taxonomy_all/99/consensus_taxonomy_7_levels SILVA_128_QIIME_release_taxonomy_all_99_consensus_taxonomy_7_levels.tax.txt
+else
+    echo "Didnt detect any expected directory"
+fi
+
+
+
+
+
diff --git a/conf/ref_databases.config b/conf/ref_databases.config
@@ -488,4 +488,41 @@ params {
             taxlevels = "D,P,C,O,F,G,S"
         }
     }
+    // Sidle reference databases
+    sidle_ref_databases {
+        'silva' {
+            title = "SILVA - Version 128"
+            file = [ "https://www.arb-silva.de/fileadmin/silva_databases/qiime/Silva_128_release.tgz" ]
+            tree_qza = [ "https://data.qiime2.org/2021.4/common/sepp-refs-silva-128.qza" ]
+            citation = "https://www.arb-silva.de/; Bokulich, N.A., Robeson, M., Dillon, M.R. bokulich-lab/RESCRIPt. Zenodo. http://doi.org/10.5281/zenodo.3891931"
+            license = "https://www.arb-silva.de/silva-license-information/"
+            fmtscript = "taxref_reformat_sidle.sh"
+            taxlevels = "D,P,C,O,F,G"
+        }
+        'silva=128' {
+            title = "SILVA - Version 128"
+            file = [ "https://www.arb-silva.de/fileadmin/silva_databases/qiime/Silva_128_release.tgz" ]
+            tree_qza = [ "https://data.qiime2.org/2021.4/common/sepp-refs-silva-128.qza" ]
+            citation = "https://www.arb-silva.de/; Bokulich, N.A., Robeson, M., Dillon, M.R. bokulich-lab/RESCRIPt. Zenodo. http://doi.org/10.5281/zenodo.3891931"
+            license = "https://www.arb-silva.de/silva-license-information/"
+            fmtscript = "taxref_reformat_sidle.sh"
+            taxlevels = "D,P,C,O,F,G"
+        }
+        'greengenes' {
+            title = "Greengenes - Version 13_8"
+            file = [ "ftp://greengenes.microbio.me/greengenes_release/gg_13_5/gg_13_8_otus.tar.gz" ]
+            tree_qza = [ "https://data.qiime2.org/2021.4/common/sepp-refs-gg-13-8.qza" ]
+            citation = "McDonald, D., Price, M., Goodrich, J. et al. An improved Greengenes taxonomy with explicit ranks for ecological and evolutionary analyses of bacteria and archaea. ISME J 6, 610–618 (2012). https://doi.org/10.1038/ismej.2011.139"
+            fmtscript = "taxref_reformat_sidle.sh"
+            taxlevels = "D,P,C,O,F,G,S"
+        }
+        'greengenes=13_8' {
+            title = "Greengenes - Version 13_8"
+            file = [ "ftp://greengenes.microbio.me/greengenes_release/gg_13_5/gg_13_8_otus.tar.gz" ]
+            tree_qza = [ "https://data.qiime2.org/2021.4/common/sepp-refs-gg-13-8.qza" ]
+            citation = "McDonald, D., Price, M., Goodrich, J. et al. An improved Greengenes taxonomy with explicit ranks for ecological and evolutionary analyses of bacteria and archaea. ISME J 6, 610–618 (2012). https://doi.org/10.1038/ismej.2011.139"
+            fmtscript = "taxref_reformat_sidle.sh"
+            taxlevels = "D,P,C,O,F,G,S"
+        }
+    }
 }
diff --git a/modules/local/format_taxonomy_sidle.nf b/modules/local/format_taxonomy_sidle.nf
@@ -0,0 +1,38 @@
+process FORMAT_TAXONOMY_SIDLE {
+    label 'process_low'
+
+    conda "conda-forge::sed=4.7"
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        'https://containers.biocontainers.pro/s3/SingImgsRepo/biocontainers/v1.2.0_cv1/biocontainers_v1.2.0_cv1.img' :
+        'docker.io/biocontainers/biocontainers:v1.2.0_cv1' }"
+
+    input:
+    path('database.tar.gz')
+    val(suffix)
+
+    output:
+    path( "*.seq.fasta" )         , emit: seq
+    path( "*.alnseq.fasta")       , emit: alnseq
+    path( "*.tax.txt")            , emit: tax
+    path( "ref_taxonomy.*.txt")   , emit: ref_tax_info
+    path "versions.yml"           , emit: versions
+
+    when:
+    task.ext.when == null || task.ext.when
+
+    script:
+    """
+    ${params.sidle_ref_databases[params.sidle_ref_taxonomy]["fmtscript"]}
+
+    #Giving out information
+    echo -e "--sidle_ref_taxonomy: ${params.sidle_ref_taxonomy}\n" >ref_taxonomy.${suffix}.txt
+    echo -e "Title: ${params.sidle_ref_databases[params.sidle_ref_taxonomy]["title"]}\n" >>ref_taxonomy.${suffix}.txt
+    echo -e "Citation: ${params.sidle_ref_databases[params.sidle_ref_taxonomy]["citation"]}\n" >>ref_taxonomy.${suffix}.txt
+    echo "All entries: ${params.sidle_ref_databases[params.sidle_ref_taxonomy]}" >>ref_taxonomy.${suffix}.txt
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        bash: \$(bash --version | sed -n 1p | sed 's/GNU bash, version //g')
+    END_VERSIONS
+    """
+}
diff --git a/nextflow.config b/nextflow.config
@@ -118,10 +118,9 @@ params {
     kraken2_assign_taxlevels = null
     kraken2_ref_tax_custom   = null
     kraken2_confidence       = 0.0
-    sidle_ref_sequences      = null
-    sidle_ref_alignedseq     = null
     sidle_ref_taxonomy       = null
-    sidle_ref_tree           = null
+    sidle_ref_tax_custom     = null
+    sidle_ref_tree_custom    = null
 
     // MultiQC options
     multiqc_config             = null
@@ -159,7 +158,7 @@ params {
     // Schema validation default options
     validationFailUnrecognisedParams = false
     validationLenientMode            = false
-    validationSchemaIgnoreParams     = 'dada_ref_databases,qiime_ref_databases,sintax_ref_databases,kraken2_ref_databases,genomes,igenomes_base'
+    validationSchemaIgnoreParams     = 'dada_ref_databases,qiime_ref_databases,sintax_ref_databases,kraken2_ref_databases,sidle_ref_databases,genomes,igenomes_base'
     validationShowHiddenParams       = false
     validate_params                  = true
 

diff --git a/nextflow_schema.json b/nextflow_schema.json
@@ -504,24 +504,25 @@
                     "minimum": 0,
                     "maximum": 1
                 },
-                "sidle_ref_sequences": {
-                    "type": "string",
-                    "help_text": "",
-                    "description": "Path to SIDLE reference taxonomy sequences (*.fasta)"
-                },
-                "sidle_ref_alignedseq": {
+                "sidle_ref_taxonomy": {
                     "type": "string",
                     "help_text": "",
-                    "description": "Path to SIDLE aligned reference taxonomy sequences (*.fasta)"
+                    "description": "Name of supported database, and optionally also version number",
+                    "enum": [
+                        "silva",
+                        "silva=128",
+                        "greengenes",
+                        "greengenes=13_8"
+                    ]
                 },
-                "sidle_ref_taxonomy": {
+                "sidle_ref_tax_custom": {
                     "type": "string",
-                    "help_text": "",
-                    "description": "Path to SIDLE reference taxonomy strings (*.txt)"
+                    "help_text": "Consider also setting `--sidle_ref_tree_custom`. Example usage: `--sidle_ref_tax_custom 'rep_set_99.fasta,rep_set_aligned_99.fasta,taxonomy_99_taxonomy.txt'`",
+                    "description": "Comma separated paths to three files: reference taxonomy sequences (*.fasta), reference taxonomy strings (*.txt)"
                 },
-                "sidle_ref_tree": {
+                "sidle_ref_tree_custom": {
                     "type": "string",
-                    "help_text": "",
+                    "help_text": "Overwrites tree chosen by `--sidle_ref_taxonomy`",
                     "description": "Path to SIDLE reference taxonomy tree (*.qza)"
                 },
                 "sintax_ref_taxonomy": {

diff --git a/subworkflows/local/sidle_wf.nf b/subworkflows/local/sidle_wf.nf
@@ -2,33 +2,44 @@
  * Training of a classifier with QIIME2
  */
 
-include { SIDLE_INDB        } from '../../modules/local/sidle_indb'
-include { SIDLE_INDBALIGNED } from '../../modules/local/sidle_indbaligned'
-include { SIDLE_DBFILT      } from '../../modules/local/sidle_dbfilt'
-include { SIDLE_IN          } from '../../modules/local/sidle_in'
-include { SIDLE_TRIM        } from '../../modules/local/sidle_trim'
-include { SIDLE_DBEXTRACT   } from '../../modules/local/sidle_dbextract'
-include { SIDLE_ALIGN       } from '../../modules/local/sidle_align'
-include { SIDLE_DBRECON     } from '../../modules/local/sidle_dbrecon'
-include { SIDLE_TABLERECON  } from '../../modules/local/sidle_tablerecon'
-include { SIDLE_TAXRECON    } from '../../modules/local/sidle_taxrecon'
-include { SIDLE_FILTTAX     } from '../../modules/local/sidle_filttax'
-include { SIDLE_SEQRECON    } from '../../modules/local/sidle_seqrecon'
-include { SIDLE_TREERECON   } from '../../modules/local/sidle_treerecon'
-
+include { FORMAT_TAXONOMY_SIDLE } from '../../modules/local/format_taxonomy_sidle'
+include { SIDLE_INDB            } from '../../modules/local/sidle_indb'
+include { SIDLE_INDBALIGNED     } from '../../modules/local/sidle_indbaligned'
+include { SIDLE_DBFILT          } from '../../modules/local/sidle_dbfilt'
+include { SIDLE_IN              } from '../../modules/local/sidle_in'
+include { SIDLE_TRIM            } from '../../modules/local/sidle_trim'
+include { SIDLE_DBEXTRACT       } from '../../modules/local/sidle_dbextract'
+include { SIDLE_ALIGN           } from '../../modules/local/sidle_align'
+include { SIDLE_DBRECON         } from '../../modules/local/sidle_dbrecon'
+include { SIDLE_TABLERECON      } from '../../modules/local/sidle_tablerecon'
+include { SIDLE_TAXRECON        } from '../../modules/local/sidle_taxrecon'
+include { SIDLE_FILTTAX         } from '../../modules/local/sidle_filttax'
+include { SIDLE_SEQRECON        } from '../../modules/local/sidle_seqrecon'
+include { SIDLE_TREERECON       } from '../../modules/local/sidle_treerecon'
 
 workflow SIDLE_WF {
     take:
     ch_asv_tables_sequences
-    ch_db_sequences
-    ch_db_alignedsequences
-    ch_db_taxonomy
+    ch_sidle_ref_taxonomy
+    val_sidle_ref_taxonomy
     ch_db_tree
 
     main:
     ch_sidle_versions = Channel.empty()
 
     // DB
+    if (!params.sidle_ref_tax_custom) {
+        //standard ref taxonomy input from conf/ref_databases.config, one tar.gz / tgz with all files
+        FORMAT_TAXONOMY_SIDLE ( ch_sidle_ref_taxonomy, val_sidle_ref_taxonomy )
+        ch_db_sequences = FORMAT_TAXONOMY_SIDLE.out.seqs
+        ch_db_alignedsequences = FORMAT_TAXONOMY_SIDLE.out.alnseq
+        ch_db_taxonomy = FORMAT_TAXONOMY_SIDLE.out.tax
+    } else {
+        //input from params.sidle_ref_tax_custom: it[0] = fasta = ch_db_sequences, it[1] = aligned fasta = ch_db_alignedsequences, it[2] = taxonomy txt = ch_db_taxonomy
+        ch_db_sequences = ch_sidle_ref_taxonomy.map{ it[0] }
+        ch_db_alignedsequences = ch_sidle_ref_taxonomy.map{ it[1] }
+        ch_db_taxonomy = ch_sidle_ref_taxonomy.map{ it[2] }
+    }
     SIDLE_INDB ( ch_db_sequences, ch_db_taxonomy )
     ch_sidle_versions = ch_sidle_versions.mix(SIDLE_INDB.out.versions)
     SIDLE_INDBALIGNED ( ch_db_alignedsequences )

diff --git a/workflows/ampliseq.nf b/workflows/ampliseq.nf
@@ -42,6 +42,28 @@ if (params.classifier) {
     ch_qiime_classifier = Channel.fromPath("${params.classifier}", checkIfExists: true)
 } else { ch_qiime_classifier = Channel.empty() }
 
+if (params.sidle_ref_tax_custom) {
+    if ("${params.sidle_ref_tax_custom}".contains(",")) {
+        sidle_ref_paths = "${params.sidle_ref_tax_custom}".split(",")
+        if (sidle_ref_paths.length != 3) {
+            error "--sidle_ref_tax_custom exately three filepaths separated by a comma (fasta, aligned fasta, taxonomy). Please review input."
+        }
+        ch_sidle_ref_taxonomy = Channel.fromPath( Arrays.asList(sidle_ref_paths), checkIfExists: true )
+    } else {
+        error "--sidle_ref_tax_custom accepts exately three filepaths separated by a comma. Please review input."
+    }
+    val_sidle_ref_taxonomy = "user"
+    ch_sidle_ref_taxonomy_tree = params.sidle_ref_tree_custom ? Channel.fromPath("${params.sidle_ref_tree_custom}", checkIfExists: true) : Channel.empty()
+} else if (params.sidle_ref_taxonomy) {
+    ch_sidle_ref_taxonomy = Channel.fromList( params.sidle_ref_databases[params.sidle_ref_taxonomy]["file"] ).map { file(it) }
+    ch_sidle_ref_taxonomy_tree = params.sidle_ref_tree_custom ? Channel.fromPath("${params.sidle_ref_tree_custom}", checkIfExists: true) : Channel.fromList( params.sidle_ref_databases[params.sidle_ref_taxonomy]["tree_qza"] ).map { file(it) }
+    val_sidle_ref_taxonomy = params.sidle_ref_taxonomy.replace('=','_').replace('.','_')
+} else {
+    ch_sidle_ref_taxonomy = Channel.empty()
+    ch_sidle_ref_taxonomy_tree = Channel.empty()
+    val_sidle_ref_taxonomy = "none"
+}
+
 if (params.dada_ref_tax_custom) {
     //custom ref taxonomy input from params.dada_ref_tax_custom & params.dada_ref_tax_custom_sp
     ch_assigntax = Channel.fromPath("${params.dada_ref_tax_custom}", checkIfExists: true)
@@ -439,10 +461,9 @@ workflow AMPLISEQ {
         // run q2-sidle
         SIDLE_WF (
             DADA2_SPLITREGIONS.out.for_sidle,
-            file( params.sidle_ref_sequences, checkIfExists: true ), //TODO: ch_sidle_ref_sequences // "gg_13_8_otus_rep_set_99_otus.fasta"
-            file( params.sidle_ref_alignedseq, checkIfExists: true ), //TODO: ch_sidle_ref_alignedseq // "gg_13_8_otus_taxonomy_99_otu_taxonomy.txt"
-            file( params.sidle_ref_taxonomy, checkIfExists: true ), //TODO: ch_sidle_ref_taxonomy // "gg_13_8_otus_taxonomy_99_otu_taxonomy.txt"
-            file( params.sidle_ref_tree, checkIfExists: true ) //TODO: ch_sidle_ref_tree // https://data.qiime2.org/2021.4/common/sepp-refs-gg-13-8.qza
+            ch_sidle_ref_taxonomy.collect(),
+            val_sidle_ref_taxonomy,
+            ch_sidle_ref_taxonomy_tree
         )
         ch_versions = ch_versions.mix(SIDLE_WF.out.versions)
     }