Added offline run flag and profile

conf/offline.config

@@ -0,0 +1,15 @@
+/*
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+    Nextflow config file for offline run.
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+*/
+
+params {
+    config_profile_name        = 'Offline'
+    config_profile_description = 'Settings for offline run'
+
+    // Other parameters
+    offline_run     = true
+    local_databases = true
+    skip_downstream = true
+}

modules/local/fetch_eggnog_group_local.nf

@@ -28,7 +28,7 @@ process FETCH_EGGNOG_GROUP_LOCAL {
     prefix = task.ext.prefix ?: meta.id
     """
     uniprotid=\$(zcat $idmap | grep \$(cat $uniprot_id) | cut -f2)
-    zcat $db | grep \$uniprotid | cut -f 5 | tr ',' '\n' | awk -F'.' '{ print \$2 }' > ${prefix}_eggnog_group_raw.txt
+    zcat $db | grep \$uniprotid | cut -f 5 | tr ',' '\\n' | awk -F'.' '{ print \$2 }' > ${prefix}_eggnog_group_raw.txt
     uniprotize_oma_online.py ${prefix}_eggnog_group_raw.txt > ${prefix}_eggnog_group.txt
     csv_adorn.py ${prefix}_eggnog_group.txt EggNOG > ${prefix}_eggnog_group.csv
 

modules/local/write_seqinfo.nf

@@ -9,6 +9,7 @@ process WRITE_SEQINFO {
 
     input:
     tuple val(meta), val(uniprot_id)
+    val offline_run
 
     output:
     tuple val(meta), path("*_id.txt"), path("*_taxid.txt"), path("*_exact.txt") , emit: seqinfo
@@ -19,10 +20,11 @@ process WRITE_SEQINFO {
 
     script:
     prefix = task.ext.prefix ?: meta.id
+    tax_command = offline_run ? "echo 'UNKNOWN' > ${prefix}_taxid.txt" : "fetch_oma_taxid_by_id.py $uniprot_id > ${prefix}_taxid.txt"
     """
     echo "${uniprot_id}" > ${prefix}_id.txt
     echo "true" > ${prefix}_exact.txt
-    fetch_oma_taxid_by_id.py $uniprot_id > ${prefix}_taxid.txt
+    $tax_command
 
     cat <<- END_VERSIONS > versions.yml
     "${task.process}":

nextflow.config

@@ -21,6 +21,7 @@ params {
 
     // Ortholog options
     use_all                    = false
+    offline_run                = false
     local_databases            = false
     skip_oma                   = false
     oma_path                   = null
@@ -202,6 +203,7 @@ profiles {
     test       { includeConfig 'conf/test.config'       }
     test_fasta { includeConfig 'conf/test_fasta.config' }
     test_full  { includeConfig 'conf/test_full.config'  }
+    offline    { includeConfig 'conf/offline.config'    }
 }
 
 // Set default registry for Apptainer, Docker, Podman and Singularity independent of -profile

nextflow_schema.json

@@ -63,6 +63,13 @@
                     "help_text": "If set to `true`, the pipeline will use local databases for the analysis.",
                     "fa_icon": "fas fa-database"
                 },
+                "offline_run": {
+                    "type": "boolean",
+                    "default": "false",
+                    "description": "Run the pipeline in offline mode. Overrides all online database flags.",
+                    "help_text": "If set to `true`, the pipeline will run in offline mode. `local_databases` must be set separately.",
+                    "fa_icon": "fas fa-database"
+                },
                 "skip_oma": {
                     "type": "boolean",
                     "default": "false",

subworkflows/local/get_orthologs.nf

@@ -28,6 +28,14 @@ workflow GET_ORTHOLOGS {
     ch_versions    = Channel.empty()
     ch_orthogroups = Channel.empty()
 
+    fasta_input = false
+    ch_samplesheet_fasta.ifEmpty {
+        fasta_input = true
+    }
+    if (fasta_input && params.offline_run) {
+        error "Offline run is currently not supported with fasta files as input."
+    }
+
     // Preprocessing - find the ID and taxid of the query sequences
     ch_samplesheet_fasta
         .map { it -> [it[0], file(it[1])] }
@@ -41,14 +49,19 @@ workflow GET_ORTHOLOGS {
     ch_versions = ch_versions.mix(IDENTIFY_SEQ_ONLINE.out.versions)
 
     WRITE_SEQINFO (
-        ch_samplesheet_query
+        ch_samplesheet_query,
+        params.offline_run
     )
 
     ch_query = IDENTIFY_SEQ_ONLINE.out.seqinfo.mix(WRITE_SEQINFO.out.seqinfo)
     ch_versions = ch_versions.mix(WRITE_SEQINFO.out.versions)
 
     // Ortholog fetching
 
+    if(params.use_all && params.offline_run) {
+        warning("Trying to use online databases in offline mode. Are you sure?") // TODO: make a warning
+    }
+
     if(params.use_all) {
         // OMA
         if (params.local_databases) {