Multivariate Analysis of Transcript Splicing (MATS)

Xing Lab, University of California, Los Angeles

Install rMATS:

Install Python 2.6.x or Python 2.7.x and corresponding versions of NumPy and SciPy

Download and install samtools (version 0.1.19)

Download and install bowtie (version 0.12.8) & tophat (version 1.4.0 or 1.4.1)

Add the Python directory to the $PATH environment variable

Add the bowtie and tophat directories to the $PATH environment variable

Add the samtools directory to the $PATH environment variable

Download rMATS

Obtain bowtie index for genome by either of the following two ways
Download pre-built bowtie indexes if using Human (hg19) or Mouse (mm9)
Build own bowtie index using bowtie-build from genome fasta sequence

Untar rMATS and bowtie indexes. For example, assuming you use pre-build bowtie indexes, unpack rMATS.3.0.9.tgz in your home directory and unpack bowtieIndexes.tgz in your bowtieIndexes directory:

cd
tar -xzf rMATS.3.0.9.tgz
cd ~/bowtieIndexes
tar -xzf bowtieIndexes.tgz

If you have questions, the FAQ page and rMATS Users Google Group are good resources.

Test rMATS:

Run testRun.sh as below to test rMATS runs properly.

cd ~/rMATS.3.0.9
./testRun.sh ~/bowtieIndexes/hg19

This two outputs can be found in the fastqTest and bamTest directories. The test run output should look like the rMATS output description.

Trim Fastq (Optional):

To trim the poor quality 3' end of reads, use the trimFastq.py script found in the bin directory.

python trimFastq.py input.fastq trimmed.fastq desired_length

cd ~/rMATS.3.0.9/
python bin/trimFastq.py testData/231EV.25K.rep-1.R1.fastq testData/trimmed.fastq 32

The above command trims 231EV.25K.rep-1.R1.fastq to 32 bp long by removing sequence from the 3' end of the reads and then saves it to trimmed.fastq.

Alternative Splicing Events

rMATS analyzes skipped exon (SE), alternative 5' splice site (A5SS), alternative 3' splice site (A3SS), mutually exclusive exons (MXE), and retained intron (RI) events. Possible alternative splicing events are identified from the RNA-Seq data and annotation of transcripts in GTF format. The following is a list of provided GTF files found in the gtf directory:

Human, Homo sapiens (Ensembl or UCSC Known Genes)

Mouse, Mus musculus (Ensembl or UCSC Known Genes)

Drosophila, Drosophila melanogaster (FlyBase)

C. Elegans, Caenorhabditis elegans (RefSeq)

Zebrafish, Danio rerio (RefSeq)

Alternatively, you can download your own transcript annotation in GTF format. However, the first column (chromosome/contig name) in the GTF must match the sequence names in your bowtie index. Use bowtie-inspect (found in the bowtie directory) to display sequence names for the bowtie index.

bowtie-inspect --names your_bowtie_index

Only use a GTF in which the chromosome/contig name (first column) matches with the above command output.

Using rMATS:

The following is a detailed description of the options used with rMATS.

Usage:

Running with fastq

python RNASeq-MATS.py -s1 rep1_1[:rep1_2][,rep2_1[:rep2_2]]* -s2 rep1_1[:rep1_2][,rep2_1[:rep2_2]]* -gtf gtfFile -bi bowtieIndexBase -o outDir -t readType -len readLength [options]*

Running with bam

python RNASeq-MATS.py -b1 s1_rep1.bam[,s1_rep2.bam]* -b2 s2.rep1.bam[,s2.rep2.bam]* -gtf gtfFile -o outDir -t readType -len readLength [options]*

Required Parameters:

-s1 rep1_1[:rep1_2][,rep2_1[:rep2_2]]*	FASTQ file(s) for the sample_1. For the paired-end data, two files must be colon separated and replicates must be in a comma separated list (Only if using fastq)
-s2 rep1_1[:rep1_2][,rep2_1[:rep2_2]]*	FASTQ file(s) for the sample_2. For the paired-end data, two files must be colon separated and replicates must be in a comma separated list (Only if using fastq)
-b1 s1_rep1.bam[,s1_rep2.bam]	Mapping results for the sample_1 in bam format. Replicates must be in a comma separated list (Only if using bam)
-b2 s2.rep1.bam[,s2.rep2.bam]	Mapping results for the sample_2 in bam format. Replicates must be in a comma separated list (Only if using bam)
-t readType	Type of read used in the analysis. readType is either 'paired' or 'single'. 'paired' is for paired-end data and 'single' is for single-end data
-len <int>	The length of each read
-gtf gtfFile	An annotation of genes and transcripts in GTF format
-bi bowtieIndexBase	The basename of the bowtie indexes (ebwt files). The base name does not include the first period. For example, use hg19 for hg19.1.ebwt. (Only if using fastq)
-o outDir	The output directory

Optional:

-a <int>	The "anchor length" used in TopHat. At least “anchor length” NT must be mapped to each end of a given junction. The default is 8
-r1 <float>[,<float>]*	The insert sizes of sample_1 data. This applies only for the paired-end data. Replicates are separated by comma. The default is 100 for each replicate
-r2 <float>[,<float>]*	The insert sizes of sample_2 data. This applies only for the paired-end data. Replicates are separated by comma. The default is 100 for each replicate
-sd1 <float>[,<float>]*	The standard deviations for the r1. Replicates are separated by comma. The default is 300 for each replicate
-sd2 <float>[,<float>]*	The standard deviations for the r2. Replicates are separated by comma. The default is 300 for each replicate
-c <float>	The cutoff splicing difference. The cutoff used in the null hypothesis test for differential splicing. The default is 0.0001 for 0.01% difference. Valid: 0 ≤ cutoff < 1
-analysis analysisType	Type of analysis to perform. analysisType is either 'P' or 'U'. 'P' is for paired analysis and 'U' is for unpaired analysis. The default is 'U'
-keepTemp	Enables rMATS to keep its temporary files generated during the run. Temporary files are generally for debugging. The default is to delete all temporary files.

Examples:

Example using fastq, performing paired analysis

python RNASeq-MATS.py -s1 testData/231ESRP.25K.rep-1.R1.fastq:testData/231ESRP.25K.rep-1.R2.fastq,testData/231ESRP.25K.rep-2.R1.fastq:testData/231ESRP.25K.rep-2.R2.fastq -s2 testData/231EV.25K.rep-1.R1.fastq:testData/231EV.25K.rep-1.R2.fastq,testData/231EV.25K.rep-2.R1.fastq:testData/231EV.25K.rep-2.R2.fastq -gtf gtf/Homo_sapiens.Ensembl.GRCh37.65.gtf -bi ~/bowtieIndexes/hg19 -o out_test -t paired -len 50 -a 8 -r1 72,75 -sd1 40,35 -r2 70,65 -sd2 48,45 -c 0.0001 -analysis P

Example using bam, performing unpaired analysis

python RNASeq-MATS.py -b1 testData/231ESRP.25K.rep-1.bam,testData/231ESRP.25K.rep-2.bam -b2 testData/231EV.25K.rep-1.bam,testData/231EV.25K.rep-2.bam -gtf gtf/Homo_sapiens.Ensembl.GRCh37.65.gtf -o bam_test -t paired -len 50 -a 8 -c 0.0001 -analysis U

Output:

All output files are in outputFolder

MATS_output: A folder that contains rMATS output of AS events. Each output file is sorted by P-values in ascending order.

AS_Event.MATS.JunctionCountOnly.txt evaluates splicing with only reads that span splicing junctions
IJC_SAMPLE_1: inclusion junction counts for SAMPLE_1, replicates are separated by comma
SJC_SAMPLE_1: skipping junction counts for SAMPLE_1, replicates are separated by comma
IJC_SAMPLE_2: inclusion junction counts for SAMPLE_2, replicates are separated by comma
SJC_SAMPLE_2: skipping junction counts for SAMPLE_2, replicates are separated by comma

AS_Event.MATS.ReadsOnTargetAndJunctionCounts.txt evaluates splicing with reads that span splicing junctions and reads on target (striped regions on home page figure)
IC_SAMPLE_1: inclusion counts for SAMPLE_1, replicates are separated by comma
SC_SAMPLE_1: skipping counts for SAMPLE_1, replicates are separated by comma
IC_SAMPLE_2: inclusion counts for SAMPLE_2, replicates are separated by comma
SC_SAMPLE_2: skipping counts for SAMPLE_2, replicates are separated by comma

Important columns contained in both types of output files
IncFormLen: length of inclusion form, used for normalization
SkipFormLen: length of skipping form, used for normalization
IncLevel1: inclusion level for SAMPLE_1 replicates (comma separated) calculated from normalized counts
IncLevel2: inclusion level for SAMPLE_2 replicates (comma separated) calculated from normalized counts
IncLevelDifference: average(IncLevel1) - average(IncLevel2)

summary.txt: A file that contains summary of statistically significant AS events and the identity of each replicate

ASEvents: A folder that contains all possible alternative splicing (AS) events derived from GTF and RNA

SAMPLE_1/REP_N: A folder that contains mapping results of sample_1, replicate N
accepted_hits.bam is the original tophat output containing both multi-mapped and uniquely mappable reads.
unique.S1.sam contains uniquely mappable reads only. rMATS uses uniquely mappable reads.

SAMPLE_2/REP_N: A folder that contains mapping results of sample_2, replicate N
accepted_hits.bam is the original tophat output containing both multi-mapped and uniquely mappable reads.
unique.S2.sam contains uniquely mappable reads only. rMATS uses uniquely mappable reads.

commands.txt: A list of key commands executed

log.RNASeq-MATS: Log file for running rMATS pipeline