Incorporating Large Language Model-Derived Information into Hypothesis Testing for Genomics

We propose strategies for incorporating the information in large language models (LLMs) into statistical hypothesis tests in genomics studies. Using gene embeddings derived from text inputs to OpenAIs GPT-3.5 model, we show that biological signals in a variety of genomics datasets reside near the principal subspace spanned by the embeddings. We then use a frequentist and Bayesian (FAB) framework to propose several hypothesis tests that are either optimal or approximately optimal with respect to prior information based on the gene embedding subspace. In four real-world genomics examples, the FAB tests guided by the LLM-derived information achieve more power than classical counterparts.