Merge branch 'rna' into d1

sanger-tol · Jan 7, 2025 · 59edad5 · 59edad5
2 parents 9d3f933 + cb720e9
commit 59edad5
Show file tree

Hide file tree

Showing 30 changed files with 32,426 additions and 591 deletions.
diff --git a/.pre-commit-config.yaml b/.pre-commit-config.yaml
@@ -11,4 +11,4 @@ repos:
     hooks:
       - id: editorconfig-checker
         alias: ec
-        exclude: .*\.hmm
+        exclude: .*(\.hmm$|.*\.cm$)
diff --git a/assets/rRNA.cm b/assets/rRNA.cm
diff --git a/bin/bin_summary.R b/bin/bin_summary.R
@@ -41,11 +41,21 @@ parser <- add_option(
 
 parser <- add_option(
     object = parser,
-    opt_str = c("-r", "--rrnas"),
+    opt_str = c("-l", "--trnas"),
     type = "character",
     action = "store",
     default = NULL,
-    help = "Comma-separated list of TSV files output by GAWK_PROKKA_SUMMARY.",
+    help = "Comma-separated list of TSV files output by GAWK_TRNASCAN_SUMMARY.",
+    metavar="filename"
+)
+
+parser <- add_option(
+    object = parser,
+    opt_str = c("-g", "--rrnas"),
+    type = "character",
+    action = "store",
+    default = NULL,
+    help = "Comma-separated list of TSV files output by BIN_RRNAS",
     metavar="filename"
 )
 
@@ -129,10 +139,8 @@ read_taxonomy <- function(file) {
     return(df)
 }
 
-read_rrnas <- function(file) {
-    df <- read_tsv(file)
-    return(df)
-}
+read_trnas <- read_tsv
+read_rrnas <- read_tsv
 
 ## Takes the arg input list and a defined input type
 ## Check if the arg has been passed, then split the string into
@@ -150,7 +158,7 @@ split_and_read <- function(input, input_type) {
 
 ## Map across all input types, read them, discard any that weren't provided
 ## and then bind them all together by bin
-input_types <- c("stats", "checkm2", "taxonomy", "rrnas")
+input_types <- c("stats", "checkm2", "taxonomy", "trnas", "rrnas")
 summary <- map(input_types, \(x) split_and_read(input, x)) |>
     discard(is.null) |>
     reduce(\(x, y) left_join(x, y, by = "bin"))

diff --git a/bin/prokka_summary.awk b/bin/prokka_summary.awk
diff --git a/bin/trnascan_summary.awk b/bin/trnascan_summary.awk
@@ -0,0 +1,24 @@
+## Summarise information about tRNA and rRNA contents
+## given by Prokka for a genome bin
+##
+## Author: Jim Downie, 2024
+
+BEGIN {
+    FS = "\t"
+    OFS = FS
+    print "bin", "total_trnas", "unique_trnas"
+}
+BEGINFILE {
+    bin = FILENAME
+    sub(".*/", "", bin)
+    sub(/\.[^\.]+$/, "", bin)
+    unique_trnas = 0
+}
+{
+    trna_arr[$5] = 1
+}
+ENDFILE {
+    for (i in trna_arr) { unique_trnas++ }
+    delete trna_arr
+    print bin, NR, unique_trnas
+}
diff --git a/conf/modules.config b/conf/modules.config
@@ -29,7 +29,17 @@ process {
         ext.prefix = { "${meta.id}_${meta.assembler}_bin3c" }
         tag        = { "${meta.id}_${meta.assembler}" }
         publishDir = [
-            path: { "${params.outdir}/bins_raw/bin3c/" },
+            path: { "${params.outdir}/bins/fasta/bin3c/" },
+            mode: params.publish_dir_mode,
+            saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
+        ]
+    }
+
+    withName: 'BIN_RRNAS' {
+        ext.prefix = { "${meta.id}_${meta.assembler}_${meta.binner}" }
+        tag = { "${meta.id}_${meta.assembler}_${meta.binner}" }
+        publishDir = [
+            path: { "${params.outdir}/bins/qc/rrna" },
             mode: params.publish_dir_mode,
             saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
         ]
@@ -38,7 +48,7 @@ process {
     withName: 'BIN_SUMMARY' {
         ext.prefix = { "${meta.id}" }
         publishDir = [
-            path: { "${params.outdir}/summary/" },
+            path: { "${params.outdir}/bins/" },
             mode: params.publish_dir_mode,
             saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
         ]
@@ -60,17 +70,17 @@ process {
         ext.prefix = { "${meta.id}_${meta.assembler}_${meta.binner}" }
         tag        = { "${meta.id}_${meta.assembler}_${meta.binner}" }
         publishDir = [
-            path: { "${params.outdir}/qc/checkm/" },
+            path: { "${params.outdir}/bins/qc/checkm/" },
             mode: params.publish_dir_mode,
             saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
         ]
     }
 
     withName: 'CONTIG2BINTOFASTA' {
-        ext.prefix = { "${meta.id}_${meta.assembler}_${binner}" }
-        tag        = { "${meta.id}_${meta.assembler}_${binner}" }
+        ext.prefix = { "${meta.id}_${meta.assembler}_${meta.binner}" }
+        tag        = { "${meta.id}_${meta.assembler}_${meta.binner}" }
         publishDir = [
-            path: { "${params.outdir}/bins_refined/${meta.binner}" },
+            path: { "${params.outdir}/bins/fasta/${meta.binner}" },
             mode: params.publish_dir_mode,
             saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
         ]
@@ -81,7 +91,7 @@ process {
         ext.prefix = { "${meta.id}_${meta.assembler}_depth" }
         tag        = { "${meta.id}_${meta.assembler}" }
         publishDir = [
-            path: { "${params.outdir}/mapping/pacbio/" },
+            path: { "${params.outdir}/${meta.assembler}/mapping/minimap2_pacbio/" },
             mode: params.publish_dir_mode,
             saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
         ]
@@ -91,11 +101,6 @@ process {
         ext.args   = { "--write_bin_evals" }
         ext.prefix = { "${meta.id}_${meta.assembler}_dastool" }
         tag        = { "${meta.id}_${meta.assembler}" }
-        publishDir = [
-            path: { "${params.outdir}/bins_refined/" },
-            mode: params.publish_dir_mode,
-            saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
-        ]
     }
 
     withName: 'FASTATOCONTIG2BIN' {
@@ -133,12 +138,12 @@ process {
         tag        = { "${meta.id}_${meta.assembler}" }
     }
 
-    withName: 'GAWK_PROKKA_SUMMARY' {
-        ext.prefix = { "${meta.id}_${meta.assembler}_${meta.binner}.prokka_summary" }
+    withName: 'GAWK_TRNASCAN_SUMMARY' {
+        ext.prefix = { "${meta.id}_${meta.assembler}_${meta.binner}.trnascan_summary" }
         tag        = { "${meta.id}_${meta.assembler}_${meta.binner}" }
         ext.suffix = { "tsv"                                         }
         publishDir = [
-            path: { "${params.outdir}/qc/prokka" },
+            path: { "${params.outdir}/bins/qc/trnascan" },
             mode: params.publish_dir_mode,
             saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
         ]
@@ -156,7 +161,7 @@ process {
         ext.prefix = { "${meta.id}_${meta.assembler}_${meta.binner}" }
         tag        = { "${meta.id}_${meta.assembler}_${meta.binner}" }
         publishDir = [
-            path: { "${params.outdir}/taxonomy/gtdbtk" },
+            path: { "${params.outdir}/bins/taxonomy/gtdbtk" },
             mode: params.publish_dir_mode,
             saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
         ]
@@ -173,6 +178,17 @@ process {
         tag        = { "${meta.id}_${meta.assembler}_${hmmfile}" }
     }
 
+    withName: 'INFERNAL_CMSEARCH' {
+        ext.args   = { "--notrunc" }
+        ext.prefix = { "${meta.id}_${meta.assembler}" }
+        tag        = { "${meta.id}_${meta.assembler}" }
+        publishDir = [
+            path: { "${params.outdir}/${meta.assembler}/qc/rrna" },
+            mode: params.publish_dir_mode,
+            saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
+        ]
+    }
+
     withName: 'MAGSCOT_MAGSCOT' {
         ext.prefix = { "${meta.id}_${meta.assembler}" }
         tag        = { "${meta.id}_${meta.assembler}" }
@@ -183,7 +199,7 @@ process {
         ext.prefix = { "${meta.id}_${meta.assembler}_maxbin2" }
         tag        = { "${meta.id}_${meta.assembler}" }
         publishDir = [
-            path: { "${params.outdir}/bins_raw/maxbin2" },
+            path: { "${params.outdir}/bins/fasta/maxbin2" },
             mode: params.publish_dir_mode,
             saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
         ]
@@ -194,14 +210,24 @@ process {
         ext.prefix = { "${meta.id}_${meta.assembler}_metabat2" }
         tag        = { "${meta.id}_${meta.assembler}" }
         publishDir = [
-            path: { "${params.outdir}/bins_raw/metabat2" },
+            path: { "${params.outdir}/bins/fasta/metabat2" },
+            mode: params.publish_dir_mode,
+            saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
+        ]
+    }
+
+    withName: 'METAMDBG_ASM' {
+        ext.args   = { "" }
+        ext.prefix = { "${meta.id}_metamdbg" }
+        publishDir = [
+            path: { "${params.outdir}/metamdbg/fasta/" },
             mode: params.publish_dir_mode,
             saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
         ]
     }
 
     withName: 'METATOR_PROCESS_INPUT_BAM' {
-        ext.prefix = { "${meta.id}_${meta.assembler}" } // metator already appends "_metator" to files
+        ext.prefix = { "${meta.id}_${meta.assembler}" }
         tag        = { "${meta.id}_${meta.assembler}" }
     }
 
@@ -210,7 +236,7 @@ process {
         ext.prefix = { "${meta.id}_${meta.assembler}" } // metator already appends "_metator" to files
         tag        = { "${meta.id}_${meta.assembler}" }
         publishDir = [
-            path: { "${params.outdir}/bins_raw/metator" },
+            path: { "${params.outdir}/bins/fasta/metator" },
             mode: params.publish_dir_mode,
             saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
         ]
@@ -222,40 +248,40 @@ process {
         ext.prefix = { "${meta2.id}_${meta2.assembler}" }
         tag        = { "${meta2.id}_${meta2.assembler}" }
         publishDir = [
-            path: { "${params.outdir}/mapping/pacbio/" },
+            path: { "${params.outdir}/${meta2.assembler}/mapping/minimap2_pacbio/" },
             mode: params.publish_dir_mode,
             saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
         ]
     }
 
-    withName: 'PROKKA' {
-        ext.args   = { "--force --metagenome --rfam --noanno --fast --rawproduct" }
-        ext.prefix = { "${meta.binid}" }
-        tag        = { "${meta.binid}" }
+    withName: 'PYRODIGAL' {
+        ext.args   = { "-p meta" }
+        ext.prefix = { "${meta.id}_${meta.assembler}" }
+        tag        = { "${meta.id}_${meta.assembler}" }
         publishDir = [
-            path: { "${params.outdir}/qc/prokka" },
+            path: { "${params.outdir}/${meta.assembler}/annotation/" },
             mode: params.publish_dir_mode,
             saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
         ]
     }
 
-    withName: 'PYRODIGAL' {
-        ext.args   = { "-p meta" }
-        ext.prefix = { "${meta.id}_${meta.assembler}" }
+    withName: 'SEQKIT_STATS_ASSEMBLIES' {
+        ext.args   = { "-b -a" }
+        ext.prefix = { "${meta.id}_${meta.assembler}.stats" }
         tag        = { "${meta.id}_${meta.assembler}" }
         publishDir = [
-            path: { "${params.outdir}/annotation/" },
+            path: { "${params.outdir}/${meta.assembler}/qc/stats/" },
             mode: params.publish_dir_mode,
             saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
         ]
     }
 
-    withName: 'SEQKIT_STATS' {
+    withName: 'SEQKIT_STATS_BINS' {
         ext.args   = { "-b -a" }
         ext.prefix = { "${meta.id}_${meta.assembler}_${meta.binner}.stats" }
         tag        = { "${meta.id}_${meta.assembler}_${meta.binner}" }
         publishDir = [
-            path: { "${params.outdir}/qc/stats/" },
+            path: { "${params.outdir}/bins/qc/stats/" },
             mode: params.publish_dir_mode,
             saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
         ]
@@ -266,7 +292,7 @@ process {
         ext.prefix = { "${meta.id}_${meta.assembler}" }
         tag        = { "${meta.id}_${meta.assembler}" }
         publishDir = [
-            path: { "${params.outdir}/mapping/hic" },
+            path: { "${params.outdir}/${meta.assembler}/mapping/bwamem2_hic/" },
             mode: params.publish_dir_mode,
             saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
         ]
@@ -277,7 +303,18 @@ process {
         ext.prefix = { "${meta.id}_${meta.assembler}_${meta.binner}.gtdb_to_ncbi" }
         tag        = { "${meta.id}_${meta.assembler}_${meta.binner}"              }
         publishDir = [
-            path: { "${params.outdir}/taxonomy/" },
+            path: { "${params.outdir}/bins/taxonomy/ncbi" },
+            mode: params.publish_dir_mode,
+            saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
+        ]
+    }
+
+    withName: 'TRNASCAN_SE' {
+        ext.args   = { "-B -Q --brief" }
+        ext.prefix = { "${meta.binid}" }
+        tag        = { "${meta.binid}" }
+        publishDir = [
+            path: { "${params.outdir}/bins/qc/trnascan_se" },
             mode: params.publish_dir_mode,
             saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
         ]

diff --git a/modules.json b/modules.json
@@ -80,12 +80,6 @@
                         "git_sha": "cf17ca47590cc578dfb47db1c2a44ef86f89976d",
                         "installed_by": ["modules"]
                     },
-                    "prokka": {
-                        "branch": "master",
-                        "git_sha": "666652151335353eef2fcd58880bcef5bc2928e1",
-                        "installed_by": ["modules"],
-                        "patch": "modules/nf-core/prokka/prokka.diff"
-                    },
                     "pyrodigal": {
                         "branch": "master",
                         "git_sha": "938e803109104e30773f76a7142442722498fef1",

diff --git a/modules/local/bin_rrnas/environment.yml b/modules/local/bin_rrnas/environment.yml
@@ -0,0 +1,7 @@
+channels:
+  - conda-forge
+  - bioconda
+dependencies:
+  - conda-forge::r-base=4.4
+  - conda-forge::r-tidyverse=2.0.0
+  - conda-forge::r-optparse=1.7.5