SpliceAI in exomiser-14

[massiddamt]

Hi, I'm trying to prioritize with exomiser with the following patosources: pathogenicitySources: [POLYPHEN, SIFT, CADD, SPLICE_AI] but no annotation for SPLICE_AI was retrieved. Could you please help me?

[VincenzoRallo]

Likewise, I'm trying to get the spliceai score in output files (both tsv and html), using it for variant priotization in pathogenicity data, but can't understand how to do it. Maybe I need to set it in application.properties? From which database is it retrieved? Thanks

[julesjacobsen]

Looks like the SpliceAI scores never made it into the database! Sorry about that. They should have come from the gnomAD annotations. @VincenzoRallo you just need to include it in the pathogenicitySources: [POLYPHEN, SIFT, CADD, SPLICE_AI] in the analysis.yml file used to configure the analysis for Exomiser.

[AlistairNWard]

Can I ask for a clarification? Do you mean that SpliceAI scores are not used in prioritizing variants at all (if the scores are not in the database, then I would assume they can't be used in the algorithm)? Or is it just that the SpliceAI scores are not included in the output files?

[julesjacobsen]

@AlistairNWard to clarify, this means that the scores are not present in the database. This means they can't be used at all so the presence of SPLICE_AI in the pathogenicitySources will make no difference to the pathogenicity score calculations.

[AlistairNWard]

Thanks @julesjacobsen, that was what we thought. We'll keep our eyes open for when there is an update to the database!

[VincenzoRallo]

Thanks all for the clarification! So if we modify the database by including the spliceAI scores, will Exomiser automatically be able to use this information in the prioritisation algorithm?

[AlistairNWard]

@julesjacobsen Do you have a estimate of when the database will be updated with SpliceAI. It will be useful for our project planning to have an idea of timeline if that's at all possible. Version 14 is working great otherwise!!

[ShrutiMarwaha]

@julesjacobsen Do you know if SpliceAI scores will be incorporated into Variant score in the next version of exomiser? Any estimated date for it? SpliceAI scores have proved to be highly useful.

[julesjacobsen]

These are now released in #562

[julesjacobsen]

A bit more info:

If you include SPLICE_AI, any SNP with a delta score >= 0.1 will be displayed and used. For synonymous variants this will likely give a better score than without SpliceAI, splice region missense variant scores will depend on how the other missense scores behave but will likely be lower than these. Intronic splice region SNPs (non-canonical splice sites) will benefit from these scores the most.