CFG Seminar: Zhongliang Zhou
PhD candidate Zhongliang Zhou from University of Georgia will present his recent work on the topic of "Unveiling the Complexity of Protein Sequence Through Effective and Interpretable Deep Learning”.
Abstract
Understanding protein-protein interactions and recognizing functional sites is crucial for deciphering protein functions and aiding drug design. This study addresses significant gaps in kinase-substrate phosphorylation prediction and protein sequence conservation through the utilization of protein language models. In the realm of kinase-substrate phosphorylation, we introduce Phosformer, an innovative deep learning model, delivering superior predictions of kinase-specific phosphosites across the entire kinome. Additionally, our interpretable Transformer model enhances generalization and transparency in kinase-peptide interaction predictions. Within the domain of sequence conservation, we present an alignment-free method that leverages protein language models to enhance the accuracy of identifying conserved functional sites in intricate protein structures. Overall, this research contributes pioneering models and methodologies that advance our comprehension of kinase-substrate interactions and protein sequence conservation, with far-reaching implications for both biological research and therapeutic applications.
View CFG's schedule
If you want to use CFG slots to share your own research, please contact Yiren Liu with any questions.
Subscribe to CFG's mailing list.
This event is sponsored by Conceptual Foundation Group (CFG) at the iSchool