Silent Failure of Annotation Weights for When variant_type != "SNV"

[kwdoyle]

When running the Gene_Centric_Coding / Gene_Centric_Noncoding functions, I recently found that functional annotation weights are silently ignored when variant_type is set to "indel" or "variant", despite no indication in the documentation suggesting this would occur.

The source of this is nested within the coding and noncoding functions, used by Gene_Centric_Coding and Gene_Centric_Noncoding

coding.R:88-117 and noncoding.R123-152: Annotation processing wrapped in if(variant_type=="SNV")

The same is true for the MetaSTAARlite functions, coding_MetaSTAARlite_worker and noncoding_MetaSTAARlite_worker

If variant_type is anything but 'SNV', the Use_annotation_weights argument essentially becomes useless, effectively forcing it to be FALSE.

[xihaoli]

Hi @kwdoyle,

Thanks for posting this. The current setup of the Gene_Centric_Coding and Gene_Centric_Noncoding functions is primarily due to the lack of functional annotations for indels. As such, if you incorporate them as weights, you may get an NA in the output of that annotation weight. To avoid these unexpected NA results, we set Use_annotation_weights = FALSE when variants in the mask (variant set) may contain indels (i.e. variant_type = "indel" or "variant").

We also have separate functions for user-defined custom masks, which can include both SNVs and indels in the mask and incorporate user-specified functional annotations as weights. Let me know if you'd like to have a copy of the custom mask functions. We will also document this setup.

Best,
Xihao