added files of template characterization-taxprofiler

Author: Daniel-VM

bu_isciii/templates/characterization/ANALYSIS/ANALYSIS02_TAXPROFILING/lablog

@@ -0,0 +1,91 @@
+# SETUP INTPUT SAMPLE SHEET
+ln -s ../00-reads .
+ln -s ../samples_id.txt .
+
+# Function to print colored text
+print_color() {
+    case "$2" in
+        "red")
+            echo -e "\e[1;31m$1\e[0m"
+            ;;
+        "green")
+            echo -e "\e[1;32m$1\e[0m"
+            ;;
+        "blue")
+            echo -e "\e[1;34m$1\e[0m"
+            ;;
+        *)
+            echo "$1"
+            ;;
+    esac
+}
+
+# Function to prompt with color
+prompt_with_color() {
+    read -p "$(print_color $1 'blue') $2" response
+}
+
+# Select whether to save trimmed reads
+trim_options=("Yes" "No")
+print_color "Do you want to save trimmed reads in outdir?" 'blue'
+select TRIMMED in "${trim_options[@]}"; do
+    if [ -n "$TRIMMED" ]; then
+        # rename trimmed
+        if [ "$TRIMMED" == "Yes" ] || [ "$TRIMMED" == "y" ]; then
+            SAVETRIMMED="true"
+        else 
+            SAVETRIMMED="false"
+        fi
+
+        break
+    else
+        print_color "Invalid input. Please select a valid option." 'red'
+    fi
+done
+print_color "Selected trimmed file option: $TRIMMED save trimmed" 'green'
+
+
+# Samples sheet setup
+echo "sample,run_accession,instrument_platform,fastq_1,fastq_2,fasta" > samplesheet.csv
+cat samples_id.txt | while read in; do
+	echo "${in},run1,ILLUMINA,00-reads/${in}_R1.fastq.gz,00-reads/${in}_R2.fastq.gz,"
+done >> samplesheet.csv
+
+scratch_dir=$(echo $PWD | sed "s/\/data\/bi\/scratch_tmp/\/scratch/g")
+
+# slurm sbatch file setup
+cat <<EOF > taxprofiler.sbatch
+#!/bin/sh
+#SBATCH --ntasks 1
+#SBATCH --cpus-per-task 2
+#SBATCH --mem 4G
+#SBATCH --time 2:00:00
+#SBATCH --partition middle_idx
+#SBATCH --output $(date '+%Y%m%d')_taxprofiler.log
+#SBATCH --chdir $scratch_dir
+
+# module load Nextflow/23.10.0 singularity
+export NXF_OPTS="-Xms500M -Xmx10G"
+
+nextflow run /data/bi/pipelines/nf-core-taxprofiler/nf-core-taxprofiler-1.1.8 \\
+    -profile singularity \\
+    -c ../../DOC/hpc_slurm_taxprofiler.config \\
+    --input samplesheet.csv \\
+    --outdir ./ \\
+    --databases ../../DOC/databasesheet.csv \\
+    --preprocessing_qc_tool fastqc \\
+    --save_preprocessed_reads ${SAVETRIMMED} \\
+    --perform_shortread_qc true \\
+    --shortread_qc_tool fastp \\
+    --perform_shortread_hostremoval true \\
+    --hostremoval_reference /data/bi/references/eukaria/homo_sapiens/hg38/NCBI/genome/GCF_000001405.40_GRCh38.p14/GCF_000001405.40_GRCh38.p14_genomic.fna.gz \\
+    --run_kraken2 true \\
+    --run_bracken true \\
+    --run_centrifuge true \\
+    --run_kaiju true \\
+    --run_metaphlan true \\
+    --run_krona true \\
+    -resume
+EOF
+
+echo "sbatch taxprofiler.sbatch" > _01_nf_taxprofiler.sh

bu_isciii/templates/characterization/ANALYSIS/lablog_taxprofiling

@@ -0,0 +1,6 @@
+mkdir -p 00-reads
+
+cd 00-reads; cat ../samples_id.txt | xargs -I % echo "ln -s ../../RAW/%_*R1*.fastq.gz %_R1.fastq.gz" | bash; cd -
+cd 00-reads; cat ../samples_id.txt | xargs -I % echo "ln -s ../../RAW/%_*R2*.fastq.gz %_R2.fastq.gz" | bash; cd -
+
+mv ANALYSIS02_TAXPROFILING $(date '+%Y%m%d')_ANALYSIS02_TAXPROFILING

bu_isciii/templates/characterization/DOC/databasesheet.csv

@@ -0,0 +1,6 @@
+tool,db_name,db_params,db_path
+kraken2,db1,,/data/bi/references/kraken/minikraken_8GB_20200312.tgz
+bracken,db2,,/data/bi/references/bracken/bracken_minikraken_8GB_20200312.tgz
+centrifuge,db3,,/data/bi/references/centrifuge/201612_centrifuge_index_p+h+v.tar.gz
+metaphlan,db4,,/data/bi/references/metaphlan/mpa_vJun23_CHOCOPhlAnSGB_20240/
+kaiju,db5,,/data/bi/references/kaiju/nr_euk_2023-05-10/

bu_isciii/templates/characterization/DOC/hpc_slurm_taxprofiler.config

@@ -0,0 +1,35 @@
+/*
+    HPC XTUTATIS CONFIGURATION
+*/
+
+singularity {
+        enabled                 = true
+        autoMounts              = true
+        singularity.cacheDir    = '/data/cnm/ratb/pipelines/singularity-images/'
+}
+
+process {
+        executor        = 'slurm'
+        queue           = 'middle_idx'
+        jobName         = { "$task.name - $task.hash" }
+        conda           = null
+
+        errorStrategy   = { task.exitStatus in ((130..145) + 104) ? 'retry' : 'finish' }
+
+    withName:'KAIJU_KAIJU' {
+                errorStrategy = { task.exitStatus in [143,137,21,1] ? 'retry' : 'finish' }
+        maxRetries    = 3
+                memory = { 72.GB * task.attempt }
+                time = { 8.h }
+        }
+}
+
+params {
+        max_memory = 376.GB
+        max_cpus = 32
+        max_time = '24.h'
+}
+
+/*
+    Custom base.config
+*/

bu_isciii/templates/characterization/RESULTS/lablog_taxprofiling_results

@@ -0,0 +1,8 @@
+DELIVERY_FOLDER="$(date '+%Y%m%d')_entrega01"
+mkdir -p $DELIVERY_FOLDER/taxprofiling
+
+# Taxprofiling service
+cd $DELIVERY_FOLDER/taxprofiling
+
+# Links to reports
+ln -s ../../../ANALYSIS/*ANALYSIS02_TAXPROFILING/results/multiqc/multiqc_report.html .