Charlie Pugh
LightNews LightNews
banner
cwjpugh.bsky.social
Charlie Pugh
@cwjpugh.bsky.social
250 followers 1.2K following 10 posts
PhD candidate - Machine Learning and Genomics @CRG.eu with @jonnyfrazer.bsky.social and @MafaldaFigDias

Posts Replies Media Videos
cwjpugh.bsky.social
This trend held across DMS assay types and mutational depth, and also on prediction of clinical variants. 8/n
We show a scatterplot of ROC-AUCs for each gene, calculated separating benign and pathogenic labelled variants with either usual or LFB fitness estimation
May 26, 2025 at 5:30 PM
cwjpugh.bsky.social
On ProteinGym, LFB provided significant improvements across model classes and sizes and we saw that larger better fit models provided better predictions in general.
proteingym.org 7/n
We show a plot of Model Size vs Mean Spearman Correlation across the DMS datasets from ProteinGym for ESM-2 and ProGen2 model families both with and without the LFB estimation.
May 26, 2025 at 5:30 PM
cwjpugh.bsky.social
We tried a simple strategy — averaging predictions over sequences under similar selective pressures to effectively reduce the impact of unwanted non-fitness related correlations — likelihood fitness bridging (LFB). 5/n
We show a schematic of the LFB estimate where by averaging over predictions for a variant applied to other related sequences, we produce an score which should be closer to the true change in fitness.
May 26, 2025 at 5:30 PM