Prediction of Gain or Loss of Function in Missense Variants

Missense variants alter a single amino acid in proteins and can cause loss-of-function (LOF) or gain-of-function (GOF). Accurately classifying these effects is essential for understanding genetic diseases and tailoring precision medicine approaches. Here, we propose the Gain and Loss of Function Dataset (GLOF), the first dataset of LOF, GOF, and Neutral missense variants annotated by specialists, curated with one of Latin America’s largest genetic diagnostics laboratories. Using embeddings from the ESM-1v model, our Random Forest model achieves state-of-the-art results without requiring complex feature engineering or multi-sequence alignment. We hope the GLOF dataset will stimulate further research into LOF/GOF prediction for personalized genomics.