From experiments to foundation models

Peptide-binding measurements are often collected for individual targets and remain scattered across projects. OpenPep aims to build a shared dataset spanning chemically diverse peptides and a broad collection of disease-relevant proteins, providing a foundation for models that can generalize across targets.

Produce targets

Cell-free protein synthesis will be used to produce a broad panel of disease-relevant proteins. The aim is to cover hundreds of targets, including proteins nominated by the scientific and clinical community, rather than concentrate measurements on a small number of proteins.

Measure binding

The target panel will be screened against libraries of synthetic peptides, including peptides built from chemical building blocks beyond the twenty natural amino acids. Affinity-selection mass spectrometry will identify candidate binders, with follow-up dose-response measurements used to characterize binding quantitatively.

The measurements are intended to form PepDB, an open database for the research community. Public releases will include file details, column descriptions, and release-specific licensing information.

Mass spectrometry is used to identify peptides recovered from binding screens.

Train foundation models

PepFM is the planned family of foundation models trained on PepDB to learn how peptides recognize proteins. The goal is to guide the selection of peptides for new experiments, including targets beyond those used in training.

Evaluation will hold proteins out of training and test predictions on those targets experimentally. This will help distinguish generalization across targets from performance on familiar examples. Results and model releases will be shared as they become available.