Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
Agentic program search was employed to discover novel nucleic acid design algorithms by iteratively proposing, executing, evaluating, and revising code, addressing the generalization failures of existing designers that exploit model-specific artifacts. AutoNovo designers achieved state-of-the-art in-silico performance on three DNA-design tasks and superior generalization across six transfer tasks to independent models, indicating a family of transferable algorithms rather than a single universal solution. This work represents a significant advancement in creating nucleic acid designers that improve the fidelity of in-silico designs to real biological function.
Novel nucleic acid design algorithms discovered via agentic program search exhibit superior cross-model generalization, outperforming existing methods on both in-silico and transfer tasks.
In-silico optimization of nucleic acids relies on predictive oracles that imperfectly capture biological function, so sequences optimized against one model may exploit model-specific artifacts and fail to generalize to independent models. Recent studies have shown that not all designers generalize equally well to independent models. In this work, we use agentic program search to iteratively propose, execute, evaluate, and revise code for nucleic-acid designers. We describe the successful strategies used to prompt the search. AutoNovo designers achieved the best observed mean among the evaluated methods on 3/3 DNA-design tasks in terms of in-silico performance, and 6/6 transfer tasks in terms of independent models. The designers come from a family of discovered nucleic acid designers rather than a single universal best designer. We plan to release the resulting designers. These results are the first steps in creating nucleic acid designers that maximize cross-model generalization, and thus improving the odds that in-silico designed sequences reflect real biology.
Google Research