AntiDIF: Accurate and Diverse Antibody Specific Inverse Folding with Discrete Diffusion

Inverse folding is an important step in currentcomputational antibody design. Recently deeplearning methods have made impressive progressin improving the sequence recovery of antibod-ies given their 3D backbone structure. However,inverse folding is often a one-to-many problem,i.e. there are multiple sequences that fold into thesame structure. Previous methods have not takeninto account the diversity between the predictedsequences for a given structure. Here we createAntiDIF an Antibody-specific discrete Diffusionmodel for Inverse Folding. Compared with state-of-the-art methods we show that AntiDIF im-proves diversity between predictions while keep-ing high sequence recovery rates. Furthermore,forward folding of the generated sequences showsgood agreement with the target 3D structure.