New DPLM-Evo model enables generative protein evolution with explicit edit predictions

Researchers have developed DPLM-Evo, a novel evolutionary discrete diffusion framework designed to better model protein evolution. Unlike previous models that used masking, DPLM-Evo explicitly predicts substitution, insertion, and deletion operations, aligning more closely with biological processes. This framework decouples latent alignment from observed sequence spaces, enabling indel-aware generation and adaptive scaffold growth. DPLM-Evo demonstrates improved sequence understanding and state-of-the-art performance in mutation effect prediction. AI

Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →

IMPACT Introduces a new method for protein sequence generation and optimization, potentially accelerating drug discovery and protein engineering.

RANK_REASON This is a research paper detailing a new model for protein evolution.

Read on arXiv cs.LG →

• DPLM-Evo
• ProteinGym
• arXiv

• paper
• model release