Merge pull request #8 from gregorysprenger/dev

Bug fixes

CHANGELOG.md

@@ -3,6 +3,20 @@
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/)
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## v2.0.2 - November 16, 2023
+
+### `Added`
+
+### `Fixed`
+
+- Added collect operators to database/reference channels to allow for all inputs to be analyzed for each process
+- Removed extra slashes '/' from run_assembly.uge-nextflow scripts
+- Updated paths for summary email when using run_assembly.uge-nextflow scripts
+
+### `Updated`
+
+### `Deprecated`
+
 ## v2.0.1 - November 16, 2023
 
 ### `Added`
@@ -24,6 +38,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - nf-core Styling
 - Allow samplesheet (XLSX, CSV, TSV) OR directory as input
 - Use SKESA instead of SPAdes for assembling contigs
+- Downsampling of input FastQ files
 - Contig taxonomic classification using GTDB-Tk
 - Intra-contig gene information using BUSCO
 - Ability to use a BUSCO config file

_run_assembly.uge-nextflow

@@ -19,7 +19,6 @@ nextflow \
   -profile ${HPC} \
   --input ${IN} \
   --outdir ${OUT} \
-  -ansi-log false \
   -N ${USER}@cdc.gov \
   -w ${OUT}/.work \
   --blast_db ${LAB_HOME}/.databases/ncbi \
@@ -59,6 +58,12 @@ while read -r line; do
       error=$(grep -A1 "java.lang" ${line}/.command.err | head -n 2 | tail -n 1)
     fi
 
+    # Drop BBDuk reformatting message
+    if [[ ${process} =~ .*REMOVE_PHIX_BBDUK$ ]] \
+      && [[ "${error}" =~ "Input is being processed as unpaired" ]]; then
+      continue
+    fi
+
     # Check if error is from file checks
     if [[ ${error} =~ .+Check[[:space:]]failed$ ]]; then
       get_previous_process_workdir=$(dirname $(grep "ln -s" ${line}/.command.run | grep "work" | awk 'END {print $(NF-1)}' ))
@@ -101,25 +106,25 @@ if [[ -f "${OUT}/pipeline_info/errors.tsv" ]]; then
 fi
 
 # Count lines in Summary.Illumina.GenomeCoverage.tab
-if [[ -f "${OUT}/qa/Summary.Illumina.GenomeCoverage.tab" ]]; then
-  num_assemblies=$(wc -l ${OUT}/qa/Summary.Illumina.GenomeCoverage.tab | awk '{print $1}')
+if [[ -f "${OUT}/Summaries/Summary.Illumina.GenomeCoverage.tab" ]]; then
+  num_assemblies=$(wc -l ${OUT}/Summaries/Summary.Illumina.GenomeCoverage.tab | awk '{print $1}')
 fi
 
 # E-mail completion status
-if [[ -f "${OUT}/qa/Summary.Illumina.GenomeCoverage.tab" ]] \
+if [[ -f "${OUT}/Summaries/Summary.Illumina.GenomeCoverage.tab" ]] \
   && [[ -f "${OUT}/pipeline_info/errors.tsv" ]]; then
   echo -e "Assembly and QA finished on $(date)\n${OUT}" | mail \
   -s "${num_assemblies} assembled $(basename "${OUT}") [HPC]" \
   -S smtp="smtpgw.cdc.gov" \
-  -a "${OUT}/qa/Summary.Illumina.GenomeCoverage.tab" \
+  -a "${OUT}/Summaries/Summary.Illumina.GenomeCoverage.tab" \
   -a "${OUT}/pipeline_info/errors.tsv" \
   "${USER}@cdc.gov"
 
-elif [[ -f "${OUT}/qa/Summary.Illumina.GenomeCoverage.tab" ]]; then
+elif [[ -f "${OUT}/Summaries/Summary.Illumina.GenomeCoverage.tab" ]]; then
   echo -e "Assembly and QA finished on $(date)\n${OUT}" | mail \
   -s "${num_assemblies} assembled $(basename "${OUT}") [HPC]" \
   -S smtp="smtpgw.cdc.gov" \
-  -a "${OUT}/qa/Summary.Illumina.GenomeCoverage.tab" \
+  -a "${OUT}/Summaries/Summary.Illumina.GenomeCoverage.tab" \
   "${USER}@cdc.gov"
 
 elif [[ -f "${OUT}/pipeline_info/errors.tsv" ]]; then

nextflow.config

@@ -11,7 +11,7 @@ manifest {
     description     = "Trim, assemble, and annotate paired end illumina reads."
     mainScript      = 'main.nf'
     nextflowVersion = '!>=22.04.3'
-    version         = '1.1.0'
+    version         = '2.0.2'
 }
 
 // Load configs

run_assembly.uge-nextflow

@@ -320,7 +320,7 @@ if [[ ${#submitted[@]} -ge 1 ]] && \
     -v SINGULARITY_TMPDIR=${SINGULARITY_TMPDIR} \
     -v SINGULARITY_CACHEDIR=${SINGULARITY_CACHEDIR} \
     -v NXF_SINGULARITY_CACHEDIR=${SINGULARITY_CACHEDIR} \
-    ${LAB_HOME}/workflows//wf-paired-end-illumina-assembly/_run_assembly.uge-nextflow
+    ${LAB_HOME}/workflows/wf-paired-end-illumina-assembly/_run_assembly.uge-nextflow
 
 elif [[ ${#submitted[@]} -ge 1 ]] && \
   [[ ${HOSTNAME%%.*} == 'rosalind01' ]] || \
@@ -340,7 +340,7 @@ elif [[ ${#submitted[@]} -ge 1 ]] && \
     -v SINGULARITY_TMPDIR=${SINGULARITY_TMPDIR} \
     -v SINGULARITY_CACHEDIR=${SINGULARITY_CACHEDIR} \
     -v NXF_SINGULARITY_CACHEDIR=${SINGULARITY_CACHEDIR} \
-    ${LAB_HOME}/workflows//wf-paired-end-illumina-assembly/_run_assembly.uge-nextflow
+    ${LAB_HOME}/workflows/wf-paired-end-illumina-assembly/_run_assembly.uge-nextflow
 
 else
   echo -e "${RED_TXT}Biolinux/Aspen/Rosalind HPC is not detected.\nSubmission cancelled. ${COLOR_OFF}"

workflows/assembly.nf

@@ -115,6 +115,7 @@ if (params.phix_reference) {
                                         error("PhiX reference file not in supported extension: .fasta, .fas, .fa, .fna")
                                     }
                             }
+                            .collect()
 } else {
     error("Path to PhiX reference not specified. Please supply a PhiX reference file in FastA format via `--phix_reference` parameter.")
 }
@@ -131,6 +132,7 @@ if (params.adapter_reference) {
                                         error("Adapter reference file not in supported extension: .fasta, .fas, .fa, .fna")
                                     }
                             }
+                            .collect()
 } else {
     error("Path to Adapter reference not specified. Please supply an adapter reference file in FastA format via `--adapter_reference` parameter.")
 }
@@ -312,7 +314,8 @@ workflow ASSEMBLY {
                                             } else {
                                                 error("Kraken requires 'database.{idx,kdb}' and 'taxonomy/{names,nodes}.dmp' files!")
                                             }
-                                        }
+                                    }
+                                    .collect()
         } else {
             error("Unsupported object given to --kraken1_db, database must be supplied as either a directory or a .tar.gz file!")
         }
@@ -350,6 +353,7 @@ workflow ASSEMBLY {
                                                 error("Kraken2 requires '{hash,opts,taxo}.k2d' files!")
                                             }
                                     }
+                                    .collect()
         } else {
             error("Unsupported object given to --kraken2_db, database must be supplied as either a directory or a .tar.gz file!")
         }
@@ -588,10 +592,10 @@ workflow ASSEMBLY {
         } else if ( ch_gtdbtk_db_file.isDirectory() ) {
             ch_db_for_gtdbtk = Channel
                                 .fromPath( "${ch_gtdbtk_db_file}/*", type: 'dir', maxDepth: 1 )
-                                .collect()
                                 .map{
                                     [ it[0].getSimpleName(), it ]
                                 }
+                                .collect()
         } else {
             error("Unsupported object given to --gtdb_db, database must be supplied as either a directory or a .tar.gz file!")
         }
@@ -634,6 +638,7 @@ workflow ASSEMBLY {
                                             db
                                         }
                                 }
+                                .collect()
         } else {
             error("Unsupported object given to --busco_db, database must be supplied as either a directory or a .tar.gz file!")
         }