Integration of Peptide Array Screening and Machine Learning for the Discovery of Peptides Directing Triangular Gold Nanoplate Formation

Peptide-mediated one-pot synthesis of gold nanoparticles (AuNPs) with diverse morphologies has been explored using tetrachloroauric acid ions at room temperature. Among these, triangular gold nanoplates (AuNPLs) are particularly notable for their unique optical properties and potential applications in photothermal therapy. In this study, we integrated peptide array screening with machine learning to identify peptides and evaluate their AuNP synthesis potential. By analyzing the color characteristics of AuNPs synthesized by each peptide on the array, we trained machine learning models to predict biomineralization efficiency based on peptide sequences and the resulting AuNP optical characteristics. This approach led to the discovery of four peptide sequences, among which the BR2–3 peptide (RWGGGIGGWN) facilitated the formation of uniform triangular AuNPLs with a yield of 33%. While the overall synthesis efficiency remained comparable to previously identified peptides, our findings broaden the scope of peptide sequences that can direct controlled AuNPL formation, offering possibilities for tailored AuNP synthesis. Over 95% of these nanoparticles (triangular and decahedral) fell within the 10–100 nm size range, making them highly suitable for drug delivery systems (DDS) and other biomedical applications. AuNPs synthesized with BR2–3 exhibited higher cellular uptake compared to those synthesized with other peptides, suggesting enhanced internalization potential. This study demonstrates the feasibility of peptide array-based, machine learning-assisted peptide screening for the biomineralization of triangular AuNPLs, emphasizing their applicability in biosensing and biomedicine.