Soft clipping vs EndToEnd for differentially alternative splicing analysis

[Ehsan Hajiramezanali]

Hi all,

I am doing alternative splicing analysis with a number of tools like rMATS, MISO, etc.

I do not know which kind of alignment is suitable for DAS analysis.

For example, rMATS is using EndToEnd alignment as default. But rMATS could extract more events when I applied soft clipped alignment results.

Can anybody help me with this?

What are the benefits of soft clipping vs EndToEnd alignment in DAS analysis?

I wonder if you could help me with this.

Thanks in advance.

Best regards

Ehsan

[Alexander Dobin]

Hi Ehsan,

the soft-clipping (local alignment) increases the number of mapped reads, since it trims poorly alignable tails.

Increasing the number of mapped reads increases the statistical power for downstream analysis.

However, you may want to check that the statistical model used in the downstream analysis is compatible with local alignment (i.e. doe not require end-to-end alignment).

Cheers

Alex