PeptiVerse platform unifies therapeutic peptide property prediction across sequence and SMILES representations

Developing therapeutic peptides is challenging due to the need to evaluate multiple developability properties beyond just binding affinity. Existing methods often fall short, lacking a unified approach for both canonical amino acid sequences and chemically modified peptides represented by SMILES. This fragmentation hinders efficient drug design, especially as chemically modified peptides become more prevalent. A platform capable of systematically assessing properties across these diverse representations is crucial to accelerate early-stage peptide therapeutic development and generative design workflows.

Researchers developed PeptiVerse, a universal therapeutic peptide property prediction platform. They leveraged the generalizability of large foundational models, trained on extensive protein and chemical data, to build the core prediction engine. The platform was designed to accept either amino acid sequences or chemically modified peptide SMILES as input. PeptiVerse's performance was evaluated across diverse property prediction tasks, demonstrating its capability to provide systematic property assessment. The platform is accessible via both a web interface and an open-source implementation to ensure rapid, accessible, and scalable analysis.

PeptiVerse successfully unifies therapeutic peptide property prediction, addressing the previous gap in systematic assessment across different representations. The platform demonstrated state-of-the-art performance across a variety of diverse property prediction tasks, indicating its robustness and accuracy. This capability extends to both standard amino acid sequences and complex chemically modified peptides described by SMILES. The integration of foundational models allows for broad applicability and generalizability, directly supporting early-stage development campaigns. > PeptiVerse delivers state-of-the-art performance across diverse property prediction tasks, unifying assessment for both canonical sequences and chemically modified peptide SMILES.