Swiss researchers have devised an easy-to-implement method for calibrating Raman-based models for analyte prediction.

This study shows that deep-learning-based co-folding models often produce physically unrealistic structures and are unsusceptible in their predictions to significant binding site modifications, ...