Peter Minary

AI for Computational Science Our research combines AI/ML, applied statistics, scientific computing, algorithmic development, and increasingly language-model-based approaches, motivated by questions in CRISPR, epigenetics, and structural biology, and informed by close collaboration with both computational and experimental laboratories. A central theme of the group is problem-driven method development: important scientific and real-world questions motivate new models and algorithms, and these methods in turn enable new forms of prediction, interpretation, and design. We draw on demanding problems in genome editing, molecular biology, and structural modelling to drive methodological innovation. Current research areas include: - AI x CRISPR: physically informed, probabilistic deep learning for CRISPR-based genome editing modalities - Epigenetics x CRISPR: understanding how epigenetic state and chromatin structure influence CRISPR on-target and off-target activity - Method development: simulation-assisted machine learning, uncertainty quantification, multimodal deep learning, language-model-based methods, and advanced optimisation and sampling techniques (e.g. hierarchical Monte Carlo) - 3D structural modelling: RNA (junctions, trees, pseudoknots), DNA (nucleosome architecture), and proteins (including TCR–pMHC complexes) - Software development: open computational tools including MOSAICS

Learning to utilize internal protein 3D nanoenvironment descriptors in predicting CRISPR–Cas9 off‑target activity Jeffrey Kelvin Mak, Artemi Bendandi, José Augusto Salim, Ivan Mazoni, Fabio Rogerio de Moraes, Luiz Borro, Florian Störtz, Walter Rocchia, Goran Neshich and Peter Minary In NAR Genomics and Bioinformatics. Vol. 7. No. 2. Pages lqaf054. 2025. LLM‑EmBEditor: Universal Base Editing Efficiency Prediction via Large Language Model Embeddings Lucas Schneider and Peter Minary In Proceedings of ICCBB 2025 (International Conference on Computational Biology and Bioinformatics). New York, NY, USA. 2025. Association for Computing Machinery. Learning to quantify uncertainty in off‑target activity for CRISPR guide RNAs Furkan Ozden and Peter Minary In Nucleic Acids Research. Vol