降级(电信)
雄激素受体
计算机科学
环境科学
计算生物学
癌症研究
化学
生物
遗传学
电信
前列腺癌
癌症
作者
Li Zhang,Renhong Sun,Xia Li,Chaowei Ren,Yahui Long,Wenjuan Tan,Xiaolong Wang,Ke Zou,Xiaobao Yang,Min Wu,Xiaoli Li,Xing Chen
出处
期刊:
[Cold Spring Harbor Laboratory]
日期:2025-03-02
被引量:6
标识
DOI:10.1101/2025.02.26.640266
摘要
ABSTRACT Proteolysis-targeting chimeras (PROTACs), a pioneering class of heterobifunctional ligands, have emerged as transformative tools in combating cancer and immune-related diseases due to their ability to target previously undruggable proteins and overcome drug resistance. Accurate assessment of the degradation potential of PROTACs is essential for advancing their therapeutic applications. However, the development of robust predictive models is hindered by the scarcity of large-scale datasets and domain-specific tools, as well as the underutilization of existing unlabeled data. To address these challenges, we developed an innovative method called AiPROTAC to predict the degradation capacity of PROTACs. In particular, AiPROTAC integrates graph augmentation, message-passing enhanced encoders and cross-attention mechanisms within a contrastive learning framework. Moreover, AiPROTAC leverages a curated specialized dataset combined with the PROTAC-DB dataset to enhance prediction accuracy. Experimental results across two datasets and six evaluation metrics demonstrate that AiPROTAC consistently outperforms state-of-the-art models. Further case studies underline its superior sensitivity and reliability. Notably, AiPROTAC facilitated the design of a novel androgen receptor (AR) degrader, PROTAC GT19, which achieved enhanced in vitro degradation compared to Bavdegalutamide (ARV-110). This advancement highlights our AiPROTAC’s potential to accelerate PROTAC-based therapeutic development and optimization.
科研通智能强力驱动
Strongly Powered by AbleSci AI