RNAseq-Pipeline

#!/bin/bash

#RNAseq Data Processing Pipeline

#This scipt performs quality control and mapping of sequencing reads. It intakes .gz file and trim 15 bases (based on fastQC) from 5' end and 5 bases from 3' end (read length of 100 bases). In the next step, it runs star mapping to the genome and generate read count files.

#SBATCH -A b2014097
#SBATCH -p core
#SBATCH -n 16
#SBATCH -t 100:00:00
#SBATCH -J RNA-Seq_Pipeline
#SBATCH -o /proj/b2014097/nobackup/Niyaz/logs/star-index.out
#SBATCH -e /proj/b2014097/nobackup/Niyaz/logs/star-index.err

module load bioinfo-tools
module load star/2.5.3a
module load TrimGalore/0.4.1
module load FastQC/0.11.5
module load cutadapt/1.13
module load Fastx/0.0.14


## FastX trimmer and FastQuality filter
## It intakes .gz file and trim 15 bases from 5' end and 5 bases from 3' end (based on fastQC)
## Then piped to do a quality filter

echo "trimming"

for x in *.gz; do

        file_input="$x"
        file_out="$x"'s'
        gunzip -c  $file_input | fastx_trimmer -Q33 -f 15 -l 95 | fastq_quality_filter -Q 33 -q 30 -p 85 -z -o $file_out
done;

echo "done"



## Index both genome and annotation file

STAR --runMode genomeGenerate --runThreadN 24 --genomeDir /proj/b2014097/nobackup/Niyaz/STAR/STAR-Index --genomeFastaFiles /proj/b2014097/nobackup/Niyaz/STAR/GRCh38.p10.genome.fa --sjdbGTFfile /proj/b2014097/nobackup/Niyaz/STAR/gencode.v27.annotation.gtf --sjdbOverhang 99



## Star mapping and counting reads

for read1 in *R1_001_val_1.fq.gz; do read2=$(echo $read1| sed 's/R1_001_val_1.fq.gz/R2_001_val_2.fq.gz/');

star --runThreadN 32 --runMode alignReads --readFilesCommand zcat --outFilterMultimapNmax 30 \
--outFilterScoreMinOverLread 0.3 --outFilterMatchNminOverLread 0.3 --outSAMtype BAM SortedByCoordinate \
--twopassMode Basic --outWigType bedGraph --outFileNamePrefix /proj/b2014097/nobackup/Niyaz/FastQ_files/STAR_Output/Mapped_$read1 --outWigStrand Unstranded \
--outSAMstrandField intronMotif --genomeLoad NoSharedMemory --genomeDir /proj/b2014097/nobackup/Niyaz/STAR/Genome/STAR-Index/ --quantMode GeneCounts --sjdbGTFfile /proj/b2014097/nobackup/Niyaz/STAR/Annotation/gencode.v27.annotation.gtf --readFilesIn $read1 $read2

done