A. E. Murphy

Toward generalizable and interpretable AI in regulatory genomics

It is suggested that progress requires reframing seq2func models as continually refined systems, in which targeted perturbation experiments, systematic evaluation and iterative model updates are tightly coupled through artificial intelligence-experiment feedback loops, enabling self-improving models that progressively deepen mechanistic understanding and more reliably support biological discovery.