rMATS v4.0.* (turbo)

Q: What are the differences between rMATS v4.0.* (turbo) and earlier versions?
A: rMATS v4.0.* (turbo) is the C/Cython version of rMATS v3.2.5 (refer to http://rnaseq-mats.sourceforge.net/index.html). The major differences between rMATS v4.0.* (turbo) and rMATS v3.2.5 are speed, memory usage and space usage. rMATS v4.0.* (turbo) achieves about 100 fold increase in speed compared to the published Python/R code. This would allow us to perform a large number of statistical tests in massive, multi-dimensional transcriptomics datasets. rMATS v4.0.* (turbo) is multi-threaded (able to make the best of multi-core system) while rMATS v3.2.5 is not. rMATS v3.2.5 provides multi-threading option to user, but it can utilize at most 1 CPU due to the limitation of Python (See Also global interpreter lock, or GIL). rMATS v4.0.* (turbo) integrated a set of new data structure and algorithm in order to achieve significant speedup, and it can fully enjoy the adventage of multi-core system.

Q: Can rMATS v4.0.* (turbo) handle replicates?
A: Yes, rMATS v4.0.* (turbo) is for replicates only.

Q: What if I don't have replicates?
A: Currently, rMATS v4.0.* (turbo) is for replicates only. It can not handle non-replicates.

Q: Can rMATS v4.0.* (turbo) handle strand-specific data?
A: Yes, rMATS v4.0.* (turbo) can handle strand-specific data.

Q: Can I run rMATS v4.0.* (turbo) with TopHat2?
A: rMATS v4.0.* (turbo) supports Tophat2 indirectly. Users can map their reads independently (using Tophat2 or any other aligners) then feed rMATS v4.0.* (turbo) with the resulting bam files. We strongly recommend that users map reads independently using their choice of aligner (including Tophat2) to reduce the rMATS running time and to preserve their own mapping procedures.

Q: Can I run rMATS v4.0.* (turbo) with STAR aligner output?
A: STAR aligner performs soft clipping by default which will generate variable read lengths. You can run STAR with "--alignEndsType EndToEnd" option to suppress soft clipping.

Q: I have problem downloading STAR indexes for human and mouse
A: Some browsers have a limit on downloadable file size. Use a different browser or download indexes directly from Linux command line.

Q: Does rMATS v4.0.2 (turbo) work with MacOS?
A: No, only Linux version is available for rMATS v4.0.2 (turbo).

Q: How do I run rMATS v4.0.* (turbo) for poorly annotated species?
A: Use cufflinks GTF files as input for rMATS

Q: How can I change the user-defined difference in isoform ratios?
A: Use the -c option. Increasing above the default 0.0001 will cause significant events to have larger differences but less events will be significant.

Q: How to install rMATS v4.0.* (turbo) and dependencies for it?
A: rMATS v4.0.* (turbo) is pre-built. No installation and configuration needed. Please follow the instruction on Documents to install dependencies.

Q: ImportError: /xxx/xxx/rmatspipeline.so: undefined symbol: PyUnicodeUCS2_DecodeUTF8
A: If you have this error, you are using a different Python interpreter (Unicode characters are stored as UCS-4.) for running your code than the one used to compile rMATS v4.0.* (turbo) (Unicode characters are stored as UCS-2.). Here are some common fixes:

Recompile Python with: ./configure --enable-unicode=ucs2 Use another build which is compiled with UCS4. This UCS4 version will always be released along with the UCS2 version.

Q: ImportError: /xxx/xxx/rmatspipeline.so: undefined symbol: PyUnicodeUCS4_DecodeUTF8
A: Please recompile Python with UCS4 or download rMATS v4.0.* (turbo) UCS2 version.

Q: OSError: Permission denied
A: Unix and Unix-like systems will not execute a program unless it is marked with permission to execute.
rMATS v4.0.* (turbo) has two binary file, rMATS_C/rMATSexe and rmatspipeline.so. In order to run/import these binary programs, you need to grant execute permission to them.

Q: Can I turn off the statistical model? I just want to get the read count and Alternative Splicing Event.
A: Yes, there is a parameter for this. You can specify --statoff in the parameter list so as to skip the statistical model. It's worth noting that the statistical model is highly-optimized and it won't take too much time to finish. Therefore you don't need to worry about the runnig time of your project if you forgot to turn statistical model off.