Feiran Li

The digital twin model is a precise virtual representation of a physical object, pioneering the latest generation of technological transformation. It can significantly enhance our understanding and intervention capabilities of complex biological systems, with broad potential applications in cell factory design, industrial fermentation, drug development, and personalized medicine.

Feiran Li's research focuses on digital life and has made significant progress. To address the bottleneck of slow experimental measurement of enzyme parameters in digital life model construction, she developed the first deep learning method for enzyme parameter prediction—DLKcat. This method accelerates the understanding of protein sequence-structure-function relationships and provides a general downstream functional characterization method for enzyme design and modification. Subsequently, based on DLKcat, she constructed a large-scale open-source enzyme database, GotEnzymes, which covers enzyme activity parameters for over 20 million enzyme-substrate pairs, characterizing a vast array of enzyme components for basic and applied biological sciences​.

She also developed multi-life process coupling modules, achieving a leap from metabolic modeling to multi-life process modeling in digital life, including detailed protein secretion modules. The number of reactions covered by the model increased from 4,000 to 37,000, providing rational design methods. Feiran proposed methods for automated model construction and iteration, pioneering model traceability and reproducibility, and providing automated methods for modeling non-model organisms, thus aiding the transition from microbial modeling to the more complex modeling of human cells​.

She is currently collaborating with companies to promote the application of existing digital life models and vertical domain large language models in metabolic engineering, medicine, and biopharmaceutical fields​.