Beyond nature's base pairs: machine learning-enabled design of DNA-stabilized silver nanoclusters.

Sequence-encoded biomolecules such as DNA and peptides are powerful programmable building blocks for nanomaterials. This paradigm is enabled by decades of prior research into how nucleic acid and amino acid sequences dictate biomolecular interactions. The properties of biomolecular materials can be significantly expanded with non-natural interactions, including metal ion coordination of nucleic acids and amino acids. However, these approaches present design challenges because it is often not well-understood how biomolecular sequence dictates such non-natural interactions. This Feature Article presents a case study in overcoming challenges in biomolecular materials with emerging approaches in data mining and machine learning for chemical design. We review progress in this area for a specific class of DNA-templated metal nanomaterials with complex sequence-to-property relationships: DNA-stabilized silver nanoclusters (AgN–DNAs) with bright, sequence-tuned fluorescence colors and promise for biophotonics applications. A brief overview of machine learning concepts is presented, and high-throughput experimental synthesis and characterization of AgN–DNAs are discussed. Then, recent progress in machine learning-guided design of DNA sequences that select for specific AgN–DNA fluorescence properties is reviewed. We conclude with emerging opportunities in machine learning-guided design and discovery of AgN–DNAs and other sequence-encoded biomolecular nanomaterials. [ABSTRACT FROM AUTHOR]