RNA-Seq pipeline script

#PRJEB7696 = GEO accession number for liver tissue samples
#PRJEB15314 = GEO accession number for muscle tissue samples

#high = high feed efficiency or low-RFI
#low = low feed efficiency or high-RFI

################### ALIGNMENT ################### 

#PRJEB15314_high_feed_samples = muscle tissue samples for high feed efficient group

# high_feed
 
for file in $(<PRJEB15314_high_feed_samples)
do
STAR --runThreadN 10 --genomeDir /home/slam02/joumana/ref_genome/release_92/star_index --readFilesIn /home/slam02/joumana/PRJEB15314/high/trimmed_paired_${file}_R1.fq /home/slam02/joumana/PRJEB15314/high/trimmed_paired_${file}_R2.fq --outFileNamePrefix /home/slam02/joumana/star_alignment/aligned_reads_muscle_high/${file} --outSAMstrandField intronMotif  --outSAMtype BAM Unsorted SortedByCoordinate --outFilterMismatchNmax 999 --outFilterMismatchNoverReadLmax 0.04 --outFilterMultimapNmax 1     
done
 
# check for uniquely mapped single-end (SE) reads
cd /home/slam02/joumana/star_alignment/aligned_reads_muscle_high
for file in $(<../PRJEB15314_high_feed_samples)
do samtools view -c -q 255 -F 0x2 ${file}Aligned.sortedByCoord.out.bam
done
 
# low_feed

#PRJEB15314_low_feed_samples = muscle tissue samples for low feed efficient group

for file in $(<PRJEB15314_low_feed_samples)
do
STAR --runThreadN 10 --genomeDir /home/slam02/joumana/ref_genome/release_92/star_index --readFilesIn /home/slam02/joumana/PRJEB15314/low/trimmed_paired_${file}_R1.fq /home/slam02/joumana/PRJEB15314/low/trimmed_paired_${file}_R2.fq --outFileNamePrefix /home/slam02/joumana/star_alignment/aligned_reads_muscle_low/${file} --outSAMstrandField intronMotif  --outSAMtype BAM Unsorted SortedByCoordinate --outFilterMismatchNmax 999 --outFilterMismatchNoverReadLmax 0.04 --outFilterMultimapNmax 1     
done
 
# check for uniquely mapped single-end (SE) reads
cd /home/slam02/joumana/star_alignment/aligned_reads_muscle_low
for file in $(<../PRJEB15314_low_feed_samples)
do samtools view -c -q 255 -F 0x2 ${file}Aligned.sortedByCoord.out.bam
done
 
#Liver tissue (PRJEB7696)

# high_feed
 
for file in $(<PRJEB7696_high_feed_samples)
do
STAR --runThreadN 10 --genomeDir /home/slam02/joumana/ref_genome/release_92/star_index --readFilesIn /home/slam02/joumana/PRJEB7696/high/trimmed_paired_${file}_R1.fq /home/slam02/joumana/PRJEB7696/high/trimmed_paired_${file}_R2.fq --outFileNamePrefix /home/slam02/joumana/star_alignment/aligned_reads_liver_high/${file} --outSAMstrandField intronMotif  --outSAMtype BAM Unsorted SortedByCoordinate --outFilterMismatchNmax 999 --outFilterMismatchNoverReadLmax 0.04 --outFilterMultimapNmax 1     
done
 
# check for uniquely mapped single-end (SE) reads
cd /home/slam02/joumana/star_alignment/aligned_reads_liver_high
for file in $(<../PRJEB7696_high_feed_samples)
do samtools view -c -q 255 -F 0x2 ${file}Aligned.sortedByCoord.out.bam
done
 
# low_feed
 
for file in $(<PRJEB7696_low_feed_samples)
do
STAR --runThreadN 10 --genomeDir /home/slam02/joumana/ref_genome/release_92/star_index --readFilesIn /home/slam02/joumana/PRJEB7696/low/trimmed_paired_${file}_R1.fq /home/slam02/joumana/PRJEB7696/low/trimmed_paired_${file}_R2.fq --outFileNamePrefix /home/slam02/joumana/star_alignment/aligned_reads_liver_low/${file} --outSAMstrandField intronMotif  --outSAMtype BAM Unsorted SortedByCoordinate --outFilterMismatchNmax 999 --outFilterMismatchNoverReadLmax 0.04 --outFilterMultimapNmax 1     
done
 
# check for uniquely mapped single-end (SE) reads
cd /home/slam02/joumana/star_alignment/aligned_reads_liver_low
for file in $(<../PRJEB7696_low_feed_samples)
do samtools view -c -q 255 -F 0x2 ${file}Aligned.sortedByCoord.out.bam
done

################### QUALITY CONTROL ################### 
 
for file in $(<PRJEB15314_high_feed_samples)
do
/home/slam02/RNA_seq/software/FastQC/fastqc ${file}_1.fastq ${file}_2.fastq -o QC/pre_trim/
done
 
for file in $(<PRJEB15314_low_feed_samples)
do
/home/slam02/RNA_seq/software/FastQC/fastqc ${file}_1.fastq ${file}_2.fastq -o QC/pre_trim/
done
 
for file in $(<PRJEB7696_high_feed_samples)
do
/home/slam02/RNA_seq/software/FastQC/fastqc ${file}_1.fastq ${file}_2.fastq -o QC/pre_trim/
done
 
for file in $(<PRJEB7696_low_feed_samples)
do
/home/slam02/RNA_seq/software/FastQC/fastqc ${file}_1.fastq ${file}_2.fastq -o QC/pre_trim/
done
 
for file in $(</home/slam02/joumana/star_alignment/PRJEB15314_high_feed_samples)
do
/home/slam02/RNA_seq/software/FastQC/fastqc trimmed_paired_${file}_R1.fq trimmed_paired_${file}_R2.fq -o QC_post_trim/
done
for file in $(</home/slam02/joumana/star_alignment/PRJEB15314_low_feed_samples)
do
/home/slam02/RNA_seq/software/FastQC/fastqc trimmed_paired_${file}_R1.fq trimmed_paired_${file}_R2.fq -o QC_post_trim/
done
 
for file in $(</home/slam02/joumana/star_alignment/PRJEB7696_high_feed_samples)
do
/home/slam02/RNA_seq/software/FastQC/fastqc trimmed_paired_${file}_R1.fq trimmed_paired_${file}_R2.fq -o QC_post_trim/
done
 
for file in $(</home/slam02/joumana/star_alignment/PRJEB7696_low_feed_samples)
do
/home/slam02/RNA_seq/software/FastQC/fastqc trimmed_paired_${file}_R1.fq trimmed_paired_${file}_R2.fq -o QC_post_trim/
done
  
################### ADD READ GROUPS ################### 

#PRJEB15314

#high

for file in $(</home/slam02/joumana/star_alignment/PRJEB15314_high_feed_samples)
do
java -jar /home/slam02/RNA_seq/software/picard.jar AddOrReplaceReadGroups \
      I=../${file}Aligned.sortedByCoord.out.bam  \
      O=${file}Aligned.sortedByCoord.RG.out.bam \
      RGID=${file} \
      RGLB=muscle \
      RGPL=illumina \
      RGPU=unit1 \
      RGSM=high
done
 
for file in ERR1619700 ERR1619701
do
java -jar /home/slam02/RNA_seq/software/picard.jar AddOrReplaceReadGroups \
      I=../${file}Aligned.sortedByCoord.out.bam  \
      O=${file}Aligned.sortedByCoord.RG.out.bam \
      RGID=${file} \
      RGLB=muscle \
      RGPL=illumina \
      RGPU=unit1 \
      RGSM=high
done
 
#low
 
for file in $(</home/slam02/joumana/star_alignment/PRJEB15314_low_feed_samples)
do
java -jar /home/slam02/RNA_seq/software/picard.jar AddOrReplaceReadGroups \
      I=../${file}Aligned.sortedByCoord.out.bam  \
      O=${file}Aligned.sortedByCoord.RG.out.bam \
      RGID=${file} \
      RGLB=muscle \
      RGPL=illumina \
      RGPU=unit1 \
      RGSM=low
done
 
#PRJEB7696

# high
 
for file in $(</home/slam02/joumana/star_alignment/PRJEB7696_high_feed_samples)
do
java -jar /home/slam02/RNA_seq/software/picard.jar AddOrReplaceReadGroups \
      I=../${file}Aligned.sortedByCoord.out.bam  \
      O=${file}Aligned.sortedByCoord.RG.out.bam \
      RGID=${file} \
      RGLB=liver \
      RGPL=illumina \
      RGPU=unit1 \
      RGSM=high
done
 
# low
 
for file in $(</home/slam02/joumana/star_alignment/PRJEB7696_low_feed_samples)
do
java -jar /home/slam02/RNA_seq/software/picard.jar AddOrReplaceReadGroups \
      I=../${file}Aligned.sortedByCoord.out.bam  \
      O=${file}Aligned.sortedByCoord.RG.out.bam \
      RGID=${file} \
      RGLB=liver \
      RGPL=illumina \
      RGPU=unit1 \
      RGSM=low
done
 
################### REMOVE DUPLICATES ################### 

# PRJEB15314
#high

for file in $(</home/slam02/joumana/star_alignment/PRJEB15314_high_feed_samples)
do
java -jar /home/slam02/RNA_seq/software/picard.jar MarkDuplicates \
I=../${file}Aligned.sortedByCoord.RG.out.bam \
O=${file}Aligned.sortedByCoord.RG.dedupped.out.bam \
REMOVE_DUPLICATES=true \
CREATE_INDEX=true \
VALIDATION_STRINGENCY=SILENT \
M=${file}_output_metrics.txt
done
 
 
#low
#sorted BAM files with RG information are located in
/home/slam02/joumana/star_alignment/aligned_reads_muscle_low/ReadGroups
# mkdir RemoveDuplicates
# cd RemoveDuplicates/
 
for file in $(</home/slam02/joumana/star_alignment/PRJEB15314_low_feed_samples)
do
java -jar /home/slam02/RNA_seq/software/picard.jar MarkDuplicates \
I=../${file}Aligned.sortedByCoord.RG.out.bam \
O=${file}Aligned.sortedByCoord.RG.dedupped.out.bam \
REMOVE_DUPLICATES=true \
CREATE_INDEX=true \
VALIDATION_STRINGENCY=SILENT \
M=${file}_output_metrics.txt
done

#PRJEB7696
# high

for file in $(</home/slam02/joumana/star_alignment/PRJEB7696_high_feed_samples)
do
java -jar /home/slam02/RNA_seq/software/picard.jar MarkDuplicates \
I=../${file}Aligned.sortedByCoord.RG.out.bam \
O=${file}Aligned.sortedByCoord.RG.dedupped.out.bam \
REMOVE_DUPLICATES=true \
CREATE_INDEX=true \
VALIDATION_STRINGENCY=SILENT \
M=${file}_output_metrics.txt
done

# high

for file in $(</home/slam02/joumana/star_alignment/PRJEB7696_low_feed_samples)
do
java -jar /home/slam02/RNA_seq/software/picard.jar MarkDuplicates \
I=../${file}Aligned.sortedByCoord.RG.out.bam \
O=${file}Aligned.sortedByCoord.RG.dedupped.out.bam \
REMOVE_DUPLICATES=true \
CREATE_INDEX=true \
VALIDATION_STRINGENCY=SILENT \
M=${file}_output_metrics.txt
done

################### MERGE BAM FILES ################### 

#high and low merged for each tissue (Approach ii)

#muscle_high_merged
nohup samtools merge muscle_high_merged.bam ERR1619692Aligned.sortedByCoord.RG.dedupped.out.bam ERR1619693Aligned.sortedByCoord.RG.dedupped.out.bam ERR1619696Aligned.sortedByCoord.RG.dedupped.out.bam ERR1619698Aligned.sortedByCoord.RG.dedupped.out.bam ERR1619699Aligned.sortedByCoord.RG.dedupped.out.bam ERR1619701Aligned.sortedByCoord.RG.dedupped.out.bam &

#muscle_low_merged
nohup samtools merge muscle_low_merged.bam ERR1619682Aligned.sortedByCoord.RG.dedupped.out.bam ERR1619684Aligned.sortedByCoord.RG.dedupped.out.bam ERR1619685Aligned.sortedByCoord.RG.dedupped.out.bam ERR1619688Aligned.sortedByCoord.RG.dedupped.out.bam ERR1619689Aligned.sortedByCoord.RG.dedupped.out.bam ERR1619691Aligned.sortedByCoord.RG.dedupped.out.bam &

#liver_high_merged
nohup samtools merge liver_high_merged.bam ERR669163Aligned.sortedByCoord.RG.dedupped.out.bam ERR669166Aligned.sortedByCoord.RG.dedupped.out.bam ERR669168Aligned.sortedByCoord.RG.dedupped.out.bam ERR669169Aligned.sortedByCoord.RG.dedupped.out.bam ERR669171Aligned.sortedByCoord.RG.dedupped.out.bam ERR669172Aligned.sortedByCoord.RG.dedupped.out.bam &

#liver_low_merged
nohup samtools merge liver_low_merged.bam ERR669174Aligned.sortedByCoord.RG.dedupped.out.bam ERR669177Aligned.sortedByCoord.RG.dedupped.out.bam ERR669178Aligned.sortedByCoord.RG.dedupped.out.bam ERR669179Aligned.sortedByCoord.RG.dedupped.out.bam ERR669181Aligned.sortedByCoord.RG.dedupped.out.bam ERR669182Aligned.sortedByCoord.RG.dedupped.out.bam &


#high and low merged for both tissues (Approach iii)

#feed_high_merged
nohup samtools merge feed_high_merged.bam ERR1619692Aligned.sortedByCoord.RG.dedupped.out.bam ERR1619693Aligned.sortedByCoord.RG.dedupped.out.bam ERR1619696Aligned.sortedByCoord.RG.dedupped.out.bam ERR1619698Aligned.sortedByCoord.RG.dedupped.out.bam ERR1619699Aligned.sortedByCoord.RG.dedupped.out.bam ERR1619701Aligned.sortedByCoord.RG.dedupped.out.bam ERR669163Aligned.sortedByCoord.RG.dedupped.out.bam ERR669166Aligned.sortedByCoord.RG.dedupped.out.bam ERR669168Aligned.sortedByCoord.RG.dedupped.out.bam ERR669169Aligned.sortedByCoord.RG.dedupped.out.bam ERR669171Aligned.sortedByCoord.RG.dedupped.out.bam ERR669172Aligned.sortedByCoord.RG.dedupped.out.bam &

#feed_low_merged
nohup samtools merge feed_low_merged.bam ERR1619682Aligned.sortedByCoord.RG.dedupped.out.bam ERR1619684Aligned.sortedByCoord.RG.dedupped.out.bam ERR1619685Aligned.sortedByCoord.RG.dedupped.out.bam ERR1619688Aligned.sortedByCoord.RG.dedupped.out.bam ERR1619689Aligned.sortedByCoord.RG.dedupped.out.bam ERR1619691Aligned.sortedByCoord.RG.dedupped.out.bam ERR669174Aligned.sortedByCoord.RG.dedupped.out.bam ERR669177Aligned.sortedByCoord.RG.dedupped.out.bam ERR669178Aligned.sortedByCoord.RG.dedupped.out.bam ERR669179Aligned.sortedByCoord.RG.dedupped.out.bam ERR669181Aligned.sortedByCoord.RG.dedupped.out.bam ERR669182Aligned.sortedByCoord.RG.dedupped.out.bam &

################### VARIANT CALLING ################### 

#PRJEB15314
 
#mpileup with both high and low files for muscle tissue
 
bcftools mpileup -E -Oua  -d10000000 -f /home/slam02/joumana/ref_genome/release_92/fasta_ref/Bos_taurus.UMD3.1.dna.toplevel.fa /home/slam02/joumana/star_alignment/aligned_reads_muscle_high/ReadGroups/RemoveDuplicates/PRJEB15314_high.merged.bam /home/slam02/joumana/star_alignment/aligned_reads_muscle_low/ReadGroups/RemoveDuplicates/PRJEB15314_low.merged.bam | bcftools call -mv -Ob -o PRJEB15314_high_and_low.merged.bcf
 
#PRJEB7696
 
#mpileup with both high and low files for liver tissue
 
bcftools mpileup -E -Oua  -d10000000 -f /home/slam02/joumana/ref_genome/release_92/fasta_ref/Bos_taurus.UMD3.1.dna.toplevel.fa /home/slam02/joumana/star_alignment/aligned_reads_liver_high/ReadGroups/RemoveDuplicates/PRJEB7696_high.merged.bam /home/slam02/joumana/star_alignment/aligned_reads_liver_low/ReadGroups/RemoveDuplicates/PRJEB7696_low.merged.bam | bcftools call -mv -Ob -o PRJEB7696_high_and_low.merged.bcf
 
#merged tissues
 
#mpileup with both high and low files for liver and muscle tissue
 
bcftools mpileup -E -Oua  -d10000000 -f /home/slam02/joumana/ref_genome/release_92/fasta_ref/Bos_taurus.UMD3.1.dna.toplevel.fa /home/slam02/joumana/star_alignment/aligned_reads_merged_tissues_bam/high/beef_feed_high.merged.bam /home/slam02/joumana/star_alignment/aligned_reads_merged_tissues_bam/low/beef_feed_low.merged.bam | bcftools call -mv -Ob -o Beef_feed_merged_high_and_low.merged.bcf