ABSTRACT Millions of tons of keratin-rich waste are disposed of annually, which could be revalorized, mitigating greenhouse emissions and pollution. Microbial keratinases offer tools for keratin waste valorization, yet they lack robustness for industrial use. In an Applied and Environmental Microbiology article, Y. Yang, Y. Luo, Y. Ding, Y. Yao, et al. (92:e01860-25, 2025, https://doi.org/10.1128/aem.01860-25 ) integrated machine learning, rational design, and empirical screening to produce boiling-resistant variants that were reusable and highly effective at keratin degradation, demonstrating the promise of deep-learning-guided enzyme engineering.
Francis E. Jenney (Tue,) studied this question.