feature: add a test version of dbcan3 to compare against dbcan2

dbcan3 and dbcan3-sub test versions, will run both if run_dbcan3
option is present.

Author: madeline-scyphers

nextflow.config

@@ -48,6 +48,7 @@ params {
     use_kegg = false
     use_kofam = false
     use_dbcan = false
+    use_dbcan3 = false
     use_camper = false
     use_fegenie = false
     use_methyl = false
@@ -115,6 +116,8 @@ params {
     dbcan_db = "${launchDir}/databases/dbcan/"
     dbcan_fam_activities = "${launchDir}/databases/dbcan/dbcan.fam-activities.tsv"
     dbcan_subfam_activities = "${launchDir}/databases/dbcan/dbcan.fam-activities.tsv"
+    dbcan3_db = "${launchDir}/databases/dbcan3"
+    dbcan3_sub_db = "${launchDir}/databases/dbcan3_sub"
     // vogdb
     vog_db = "${launchDir}/databases/vogdb/"
     vog_list = "${launchDir}/databases/vogdb/vog_annotations_latest.tsv.gz"
@@ -172,7 +175,7 @@ params {
     // Not the limit to the total resources available to the pipeline
     // Up to queue_size processes can run in parallel, of various sizes
     tiny_cpus_limit = 1
-    small_cpus_limit = 2
+    small_cpus_limit = 4
     medium_cpus_limit = 6
     big_cpus_limit = 12
     huge_cpus_limit = 24

nextflow_schema.json

@@ -146,6 +146,10 @@
                     "type": "boolean",
                     "description": "Use the DBCan database for annotation."
                 },
+                "use_dbcan3": {
+                    "type": "boolean",
+                    "description": "Use the experimental DBCan3 databases for annotation."
+                },
                 "use_fegenie": {
                     "type": "boolean",
                     "description": "Use the FeGenie database for annotation."
@@ -376,6 +380,16 @@
                     "default": "${launchDir}/databases/dbcan/dbcan.fam-activities.tsv",
                     "hidden": true
                 },
+                "dbcan3_db": {
+                    "type": "string",
+                    "default": "${launchDir}/databases/dbcan3/",
+                    "hidden": true
+                },
+                "dbcan3_sub_db": {
+                    "type": "string",
+                    "default": "${launchDir}/databases/dbcan3_sub/",
+                    "hidden": true
+                },
                 "vog_db": {
                     "type": "string",
                     "default": "${launchDir}/databases/vog/",

subworkflows/local/annotate.nf

@@ -18,6 +18,7 @@ workflow ANNOTATE {
     use_kegg
     use_kofam
     use_dbcan
+    use_dbcan3
     use_camper
     use_fegenie
     use_methyl
@@ -107,6 +108,7 @@ workflow ANNOTATE {
             use_kegg,
             use_kofam,
             use_dbcan,
+            use_dbcan3,
             use_camper,
             use_fegenie,
             use_methyl,

subworkflows/local/db_search.nf

@@ -32,6 +32,8 @@ include { ADD_SQL_DESCRIPTIONS as SQL_DBCAN             } from "../../modules/lo
 
 include { HMM_SEARCH as HMM_SEARCH_KOFAM                } from "../../modules/local/annotate/hmmsearch.nf"
 include { HMM_SEARCH as HMM_SEARCH_DBCAN                } from "../../modules/local/annotate/hmmsearch.nf"
+include { HMM_SEARCH as HMM_SEARCH_DBCAN3                } from "../../modules/local/annotate/hmmsearch.nf"
+include { HMM_SEARCH as HMM_SEARCH_DBCAN3_SUB                } from "../../modules/local/annotate/hmmsearch.nf"
 include { HMM_SEARCH as HMM_SEARCH_VOG                  } from "../../modules/local/annotate/hmmsearch.nf"
 include { HMM_SEARCH as HMM_SEARCH_CAMPER               } from "../../modules/local/annotate/hmmsearch.nf"
 include { HMM_SEARCH as HMM_SEARCH_CANTHYD              } from "../../modules/local/annotate/hmmsearch.nf"
@@ -53,6 +55,7 @@ workflow DB_SEARCH {
     use_kegg
     use_kofam
     use_dbcan
+    use_dbcan3
     use_camper
     use_fegenie
     use_methyl
@@ -70,6 +73,7 @@ workflow DB_SEARCH {
         use_kegg,
         use_kofam,
         use_dbcan,
+        use_dbcan3,
         use_camper,
         use_fegenie,
         use_methyl,
@@ -94,6 +98,8 @@ workflow DB_SEARCH {
 
     kegg_name = "kegg"
     dbcan_name = "dbcan"
+    dbcan3_name = "dbcan3"
+    dbcan3_sub_name = "dbcan3_sub"
     kofam_name = "kofam"
     merops_name = "merops"
     viral_name = "viral"
@@ -170,6 +176,32 @@ workflow DB_SEARCH {
         ch_dbcan_formatted = SQL_DBCAN.out.sql_formatted_hits
         formattedOutputChannels = formattedOutputChannels.mix(ch_dbcan_formatted)
     }
+    // dbCAN3 annotation
+    if  (use_dbcan3) {
+        ch_combined_proteins_locs = ch_called_proteins.join(ch_gene_locs)
+        HMM_SEARCH_DBCAN3 (
+            ch_combined_proteins_locs,
+            params.dbcan_e_value,
+            DB_CHANNEL_SETUP.out.ch_dbcan3_db,
+            default_sheet,
+            false,
+            dbcan3_name
+            )
+        ch_dbcan3_formatted = HMM_SEARCH_DBCAN3.out.formatted_hits
+        formattedOutputChannels = formattedOutputChannels.mix(ch_dbcan3_formatted)
+
+
+        HMM_SEARCH_DBCAN3_SUB (
+            ch_combined_proteins_locs,
+            params.dbcan_e_value,
+            DB_CHANNEL_SETUP.out.ch_dbcan3_sub_db,
+            default_sheet,
+            false,
+            dbcan3_sub_name
+            )
+        ch_dbcan3_sub_formatted = HMM_SEARCH_DBCAN3_SUB.out.formatted_hits
+        formattedOutputChannels = formattedOutputChannels.mix(ch_dbcan3_sub_formatted)
+    }
     // CAMPER annotation
     if (use_camper) {
         // HMM
@@ -329,6 +361,7 @@ workflow DB_CHANNEL_SETUP {
     use_kegg
     use_kofam
     use_dbcan
+    use_dbcan3
     use_camper
     use_fegenie
     use_methyl
@@ -377,6 +410,11 @@ workflow DB_CHANNEL_SETUP {
         ch_dbcan_db = file(params.dbcan_db).exists() ? file(params.dbcan_db) : error("Error: If using --annotate, you must supply prebuilt databases. DBCAN database file not found at ${params.dbcan_db}")
     }
 
+    if (use_dbcan3) {
+        ch_dbcan3_db = file(params.dbcan3_db).exists() ? file(params.dbcan3_db) : error("Error: If using --annotate, you must supply prebuilt databases. DBCAN3 database file not found at ${params.dbcan3_db}")
+        ch_dbcan3_sub_db = file(params.dbcan3_sub_db).exists() ? file(params.dbcan3_sub_db) : error("Error: If using --annotate, you must supply prebuilt databases. DBCAN3 sub database file not found at ${params.dbcan3_sub_db}")
+    }
+
     if (use_camper) {
         ch_camper_hmm_db = file(params.camper_hmm_db).exists() ? file(params.camper_hmm_db) : error("Error: If using --annotate, you must supply prebuilt databases. CAMPER HMM database file not found at ${params.camper_hmm_db}")
         ch_camper_mmseqs_db = file(params.camper_mmseqs_db).exists() ? file(params.camper_mmseqs_db) : error("Error: If using --annotate, you must supply prebuilt databases. CAMPER MMseqs2 database file not found at ${params.camper_mmseqs_db}")
@@ -440,6 +478,8 @@ workflow DB_CHANNEL_SETUP {
     ch_kegg_db
     ch_kofam_db
     ch_dbcan_db
+    ch_dbcan3_db
+    ch_dbcan3_sub_db
     ch_camper_hmm_db
     ch_camper_mmseqs_db
     ch_camper_mmseqs_list

workflows/dram.nf

@@ -90,6 +90,7 @@ workflow DRAM {
         use_kegg = getDBFlag(anno_dbs, 'kegg', value_for_all)
         use_kofam = getDBFlag(anno_dbs, 'kofam', value_for_all)
         use_dbcan = getDBFlag(anno_dbs, 'dbcan', value_for_all)
+        use_dbcan3 = getDBFlag(anno_dbs, 'dbcan3', value_for_all)
         use_camper = getDBFlag(anno_dbs, 'camper', value_for_all)
         use_fegenie = getDBFlag(anno_dbs, 'fegenie', value_for_all)
         use_methyl = getDBFlag(anno_dbs, 'methyl', value_for_all)
@@ -230,6 +231,7 @@ workflow DRAM {
             use_kegg,
             use_kofam,
             use_dbcan,
+            use_dbcan3,
             use_camper,
             use_fegenie,
             use_methyl,