Our work integrates artificial intelligence (AI), among other data science techniques, to study amyloid-forming proteins, amyloid polymorphs, protein aggregation, and neurodegenerative diseases. Our adaptable methods advance precision medicine across various diseases. Our research extends to intrinsically disordered and liquid-liquid phase separation (LLPS) proteins within membrane-less organelles (MLOs), focusing on the process itself, as well as on causes of protein dysfunction such as disease-related mutations. Using protein language models (pLMs), we generate embeddings for wild-type and mutated sequences to identify novel disease-causing mutations in LLPS proteins.
We are currently focused on interpretability: opening the “black box” of pLMs, attention layers, and embeddings to extract meaningful “hidden” biological and biomedical knowledge, linking them to diseases, mutations, and other disease-relevant features