Brief tutorial for using rMATS to conduct a pairwise differential splicing analysis

rMATS_splicing.md

In order to perform a pairwise splicing analysis, sorted bam files per sample are needed as well as reference annotation in the form of a GTF file

###Download rMATS Download rMATS at http://rnaseq-mats.sourceforge.net/download.html

###Running rMATS Via python, you can run rMATs using the following example command:

python /path/to/rMATS/directory/RNASeq-MATS.py -b1 conditon1_replicate_1.bam,condition_1_replicate_2.bam -b2 condition_2_replicate_1.bam,condition_2_replicate_2.bam -gtf /path/to/gtf/file -o /path/to/results/directory -t single -len {read length}

Note:

• -t should be changed to paired for paired-end reads
• -b1 & -b2 are each a comma separated list of replicates per condition
• Results will be in the directory specified by the -o flag in a subdirectory called MATS_output. Each splicing event p-values and inclusion difference can be found in files with the suffix .ReadsOnTargetAndJunctionCounts.txt. For example, SE.MATS.ReadsOnTargetAndJunctionCounts.txt is the skipped exon result

###Visualizing Results Installation Visualizations can be made in the form of Miso's Sashimi plots using a tool called rmats2sashimiplot. First, clone the git repository from https://github.com/Xinglab/rmats2sashimiplot.git, cd into the directory and perform python setup.py install

NOTE: Miso must be installed and in your path

Selecting Splice sites Since we are only visualizing a splicing event for a single gene, we need to run the command once per gene per splicing event. This requires greping the gene from the splicng event result file. For example, visualizing a gene's skipped exon events:

grep {gene name or ensembl id} SE.ReadsOnTargetAndJunctionCounts.txt > gene_SE.txt

Then:

/rmats/path/rmats2sashimiplot -b1 conditon1_replicate_1.bam,condition_1_replicate_2.bam -b2 condition_2_replicate_1.bam,condition_2_replicate_2.bam -e gene_SE.txt -t SE -l1 Condition_1_name -l2 Condition_2_name -o /path/to/output/directory

• The -t flag is the splicing event you are investigating (SE, MXE, RI, etc...see rMATs documentation for an explanation of these)

• The l1 and l2 flags are condition label names

• The visualizations will be saved as PDFs

NOTE: unfortuantely rmats2sashimiplot requires the same number of replicates between condition 1 and condition 2. If you do not, have the same number of replicates, then it's imperative to merge all replicates to one bam file and perform the visualization