zero mapped rates uisng rmats

[wangduo-ux]

When I use STAR alone to align reads from two Fastq files, the alignment rate reaches 90%. However, when I use rMATS, the alignment rate is 0%, resulting in no output of any alternative splicing events.

[wangduo-ux]

After examining the Log.final.out file, >90% of reads were unmapped due to too many mismatches as follows:
UNMAPPED READS:
Number of reads unmapped: too many mismatches | 187596289
% of reads unmapped: too many mismatches | 95.99%
Number of reads unmapped: too short | 6937531
% of reads unmapped: too short | 3.55%
Number of reads unmapped: other | 898527
% of reads unmapped: other | 0.46%

However, when I used STAR for alignment, the unique mapped rates reached 90%.

[wangduo-ux]

Hi, Dr EricKutschera,

I have found the reasons but I still do not know how to address them.

STAR was run with the default argument "--outFilterMismatchNmax 10 ", however, STAR in rmats was run with the argument "--outFilterMismatchNmax 3", resulting in a high proportion of reads flagged as a mismatch. Thus, only <1% of reads were mapped to the genome, and few AS events were detected.

[EricKutschera]

You can align the reads outside of rMATS and then use --b1/--b2: https://github.com/Xinglab/rmats-turbo/tree/v4.3.0?tab=readme-ov-file#starting-with-bam-files

If you still want to use --s1/--s2 then you can edit the STAR parameters in https://github.com/Xinglab/rmats-turbo/blob/v4.3.0/rmats.py#L66