PSI calculation

[Swethaa NG]

Hi,

I would like to calculate PSI values for each exon of a gene in a bunch of samples. Is it possible using the Psi quantifier?

Please help, thanks!

kind regards,

Swethaa

[San Jewell]

Hi Swethaa,

Majiq quantifies PSI in terms of LSVs (local splicing variations), please see here: https://biociphers.bitbucket.io/majiq/lsv.html

Basically, this means that there is not a single defined PSI value for an exon (unless, for example, it is an alternate start/end exon, with a single LSV coming in/out of it). Because of this basic definition of Majiq's quantifications, if you want to define a PSI per exon, you'll need to determine which combination of metrics will work for your thresholds based on the LSV PSI values related to that exon.

I'm going to also relay this question to my lab to see if anyone can site more specific input or ideas on this question.

Thanks!

-San

[Swethaa NG]

Thanks for your prompt response San, appreciate it!

Ohhh okay, I understand more about the quantification of LSVs now. 

Yes, I wanted to know how to modify the thresholds or use specific metrics for the LSV PSI values related to each exon.. Maybe if I --disable-denovo and obtain the PSI values for all the LSVs spanning from & to each exon? Then for a specific gene and sample of interest, I acquire the LSV PSI values related to each exon?

And then compare those values with different sample groups?

Maybe I am describing the basic PSI quantification or deltapsi comparison here? Right?

Sorry for the confusion.. Please bear with me.

Thank you,

Swethaa

[bsl...@seas.upenn.edu]

Hi Swethaa, I'm writing to say that the lab discussed your question and we agree with San-- with MAJIQ, it's best to think of PSI for LSVs, not overall exon inclusion. 

For your analysis, one approach would be to use MAJIQ DeltaPsi or HET (comparative quantifiers) to compare two groups of RNA-seq experiments. The results will indicate whether there is differential splicing at each exon-- in the sense of some LSV junction going to or from that exon being different between the groups. Would that address your use-case?

Please let us know if you have additional questions.

Barry

[Swethaa NG]

Hi Barry,

Thanks for your message. Yes, I agree! 

Any significant differential splicing would highlight what I am looking for. I will proceed and see what I get. :)

Thanks again to the team!

Kind regards,

Swethaa