Rocío Mercado

SCARSE is a machine learning framework that improves peptide engineering by combining ESM2 embeddings with Gaussian process regression, outperforming baselines in low-data regimes and enabling efficient active learning.

This study investigates the conformational properties of 20 PROTACs to determine factors influencing their passive permeability, finding that compactness alone is not sufficient and introducing a new metric, maximum ring size.

This paper introduces CELLULAR, a contrastive learning model for scRNA-Seq data that improves embedding generalization and cell annotation performance across multiple datasets.