I want to investigate variant scores for a specific regions of the genomic sequence. Still, the CenterMaskScorer doesn’t support this and only scores around the variant nucleotide (as its name suggests). The problem is that the GeneMaskScorer family doesn’t offer a versatile aggregation method as in CenterMaskScorer. Therefore, I think it would be helpful to generalize CenterMaskScorer to MaskScorer, allowing the user to specify list of intervals to include or pass an integer for width, as before.
Hello,
Thank you so much for taking the time to share your ideas for AlphaGenome. We’ll keep you in the loop on any new developments in this forum and through our mailing list – if you haven’t subscribed yet, it’s easy to do on your settings page.
and as a quick thought - we also hope that when our model is released, this will help give you more flexibility to make these tweaks!
1 Like
if you’re passing multiple variants in at once (e.g. in silico mutagenesis), this will inevitably require multiple forward passes.
if, however, you’re interested in effects of a single variant at multiple loci, this would probably be a simple extension of the intervals feature I requested a while back (as a gentle nudge of its relevance, haha) ~
Could you please elaborate… what do you mean by multiple forward passes?