Why AI Alone Will Not Solve Oligonucleotide Discovery

By Justyna Lisowska and Ming Wang, Genedata

This content is brought to you by Genedata, a Danaher Operating Company.

AI can narrow the combinatorial search space in oligonucleotide design, but faster predictions are not the same as biological insight. The constraint is the data. Literature and patent sources vary in experimental conditions, rarely report failed experiments, and often leave safety information incomplete. Automated extraction adds its own risk of metadata errors, and larger datasets do not automatically improve predictions. This article makes the case for generating data on purpose: what separates high-value datasets from merely large ones, how systematic screening supports predictability, and why fragmented data across systems and teams limits what models can learn.

Read on to see what the argument means for your discovery workflow.