search

gagneurlab / MMSplice_MTSplice

Tissue-specific variant effect predictions on splicing
MIT License
39 stars 21 forks source link

Identifying the predicted types of splicing events #51

Open bioinfo-K1 opened 2 years ago

bioinfo-K1 commented 2 years ago

Hi, thanks for developing and maintaining this project.

I would like to classify the predicted impact based on the type of AS event which is promoted or suppressed due to the mutation. More specifically, for a mutation over an exon, identify whether the 5'end or it's 3'end will be spliced by a mutation, or if an exon skipping event is predicted. I can assume that using the donor and acceptor scores could help this this case. But the detail of those scores is not clear from the README file. Could you please further elaborate on those scores? And on how I could use the ref and alt values to classify the AS event?

s6juncheng commented 2 years ago

Thanks for using our tool. We are predicting delta_logit_psi, which is the logit PSI score change for exon skipping events. Alternative 3'end / 5'end events can also be seen as exon skipping events (skipping the shorter/longer exon for the longer/shorter exon). Therefore you can make the score into predicting alternative 3'end / 5'end. However, this version of the model is not doing this automatically.

@MuhammedHasan could you comment on predicting psi5 / psi3, whether anything is ready to be shared?

bioinfo-K1 commented 2 years ago

Thanks for your response. Have can I identify whether 3'end or 5'end of an exon is spliced out by the scores? Do you mean I would use the donor and acceptor scores to come up with such a score?