Molecular render representing ProSST

ProSST

Structure-aware mutation effect prediction

ProSST predicts mutation effects using a structure-aware language model that encodes local structural environments from predicted or experimental backbones. It improves zero-shot predictions of pathogenicity and fitness effects over sequence-only models. Apply to variant prioritization and protein engineering screens.

Read the paper

ProSST is being onboarded — request access and be first in line.

At a glance

Input
Structure and variants (PDB, VCF)
Output
Mutation effect scores (CSV)
Developed by
AI4Protein Group
Published
Li et al., NeurIPS 2024 · 2024
#mutation-effects#structure-aware#variants