RNAseq analysis pipeline

Last updated: Nov 2017

Our current preferred pipeline involves STAR/Salmon alignment and quantification followed by DESeq2 for differential expression analysis. (Differential expression may be changing to Sleuth in coming months)

Preamble

While the code below will get you going on your RNAseq analysis, I would greatly encourage looking at program documentation or tutorials for the software used.

STAR documentation
Salmon documentation
DESeq2 documentation
DESeq2 tutorial

Building indexes

Before aligning to the genome, you need to generate the indices for the genome. This has been done for mouse and human, and if you are aligning to those species, you can skip this section.

The command looks something like this (untested):

/programs/STAR_2.4.2a/STAR \
--runMode genomeGenerate \
--runThreadN 8 \
--genomeFastaFiles /proj/seth_lab/projects/genome/star_salmon_dev/mm9/mm9.fa \
--genomeDir /proj/seth_lab/projects/genome/star_salmon_dev/mm9 \
--sjdbGTFfile /proj/seth_lab/projects/genome/star_salmon_dev/mm9/mm9.gtf \
--sjdbOverhang 100

Running STAR

The STAR_multisample_single.sh is for single-read sequencing, whereas the STAR_multisample_paired.sh is for paired-end sequencing. Both are very similar, and you can look at both to see the differences, but below is the STAR_multisample_paired.sh script:

#!/bin/bash

# Parameters that need to be changed
genome=/home/pr46_0001/projects/genome/gencode.vM10

# Make output directories
rmdir STAR_OUT
rmdir SALMON_OUT
rmdir STAR_Salmon_logs

mkdir STAR_OUT
mkdir SALMON_OUT
mkdir STAR_Salmon_logs

# Loop to submit samples to STAR
for fastq in *.fastq.gz; do

# Getting information for job submission
  READ1=$fastq
  DIR=$(dirname $fastq)
  SAMP=${fastq//fastq.gz/}
  OUT_DIR=${DIR}/STAR_OUT/$SAMP
  LOGS=${DIR}/STAR_Salmon_logs

# Print out information about job submission
  echo Base name is: $SAMP
  echo Read 1 is: $READ1
  echo Output directory is: $OUT_DIR
  echo Logs are in: $LOGS

# Job submission
/programs/STAR_2.4.2a/STAR \
  --runThreadN 8 \
  --genomeDir $genome \
  --readFilesIn $READ1 \
  --outFilterMismatchNmax 20 \
  --outSAMtype BAM Unsorted \
  --quantMode TranscriptomeSAM \
  --outFileNamePrefix STAR_OUT/$SAMP \
  --outSAMunmapped Within \
  --readFilesCommand zcat \
  > STAR_Salmon_logs/${SAMP}.out

 echo
done

To run:

1. Move the STAR_multisubmission script to directory containing sequencing files

2. Make sure that your fastq files end in the extension .fastq.gz. Our STAR command expects our fastqs to be gzipped, so if you don't have gzipped fastqs, run for x in *.fastq; do gzip $x; done

3. Edit the genome location line for the species you are aligning to. In the example above we are using mouse, but we have human set up as well.
   mouse location: /home/pr46_0001/projects/genome/gencode.vM10
   human location: /home/pr46_0001/projects/genome/GRCh38.p7

4. Run STAR_multisubmission_(single or paired).sh (eg for single):

bash STAR_multisubmission_single.sh

Running Salmon

The Salmon script can infer whether single or paired-end reads were used, so the same script can be used. The script looks like this:

#!/bin/bash

# Parameters that need to be changed
transcriptome=/home/pr46_0001/projects/genome/gencode.vM10/GRCm38.p4.transcripts.fa
LOGS=STAR_Salmon_logs

for alignment in STAR_OUT/*Aligned.toTranscriptome.out.bam; do

# Get info for job submission
  NAME=$(basename $alignment Aligned.toTranscriptome.out.bam)
  OUT_DIR=SALMON_OUT/${NAME}

# Print info for job submission
  echo Base name is: $NAME
  echo Output directory is: $OUT_DIR

# Job submission (The -l will have to be changed according to type of sample prep)
  /home/pr46_0001/shared/bin/salmon quant \
  -t $transcriptome \
  -l A \
  -p 8 \
  -a STAR_OUT/${NAME}Aligned.toTranscriptome.out.bam \
  -o $OUT_DIR

  echo
done

To run:

1. Move Salmon_MultiSampleSubmission.sh to directory containing fastqs

2. Edit the location transcriptome fasta. We are using mouse, but human is also available.
   mouse transcriptome: /home/pr46_0001/projects/genome/gencode.vM10/GRCm38.p4.transcripts.fa
   human transcriptome: /home/pr46_0001/projects/genome/GRCh38.p7/gencode.v25.transcripts.fa

3. Run SALMON_MultiSampleSubmission.sh

bash SALMON_MultiSampleSubmission.sh

Once STAR/Salmon is done running, we have our counts. We can now collect the results and transfer the counts to our local machine to run the differential analysis.

To collect the results, first add the directory of the lab scripts to the system path by typing this:

PATH=$PATH:/home/pr46_0001/shared/bin

In the directory containing the sequencing fastqs, run:

RNAseq_final.py

This will create a results directory that contains our counts, the STAR mapping stats, and the Salmon mapping stats.

We can now move the results directory to our local computer for DESeq analysis. The DESeq script is commented with further instructions.

If anything is unclear, email me with your questions. -Matt