New Gospel of drug design: Tencent, together with China University of science and technology and Zhejiang University, developed an adaptive graph learning method to predict molecular interactions and

Improving the efficiency of drug discovery is the core and long-term challenge of drug discovery . So , Many graph learning methods have been developed to quickly 、 Search for potential drug candidates at low cost . in fact , The pursuit of high predictive performance on a limited number of datasets has defined their architecture and hyperparameters , They lose their advantage in reusing new data generated in drug discovery .

Tencent quantum lab and China University of science and technology 、 Team cooperation with Zhejiang University , A flexible method is proposed , It can adapt to any data set and make accurate predictions . The proposed method uses an adaptive pipeline to learn from the data set and output the predictor . Without any human intervention , This method is implemented on all test data sets , Better prediction performance than traditional methods based on manually designed neural architecture and other fixed items .

Besides , The researchers found that the proposed method is more robust than the traditional method , And can provide meaningful interpretability . In view of the above , The proposed method can be used as a reliable method to predict with high adaptability 、 performance 、 Robust and interpretable molecular interactions and properties . This work will help researchers design better drugs more efficiently .

The research 「 An adaptive graph learning method for automated molecular interactions and properties predictions」 entitled , On 2022 year 6 month 23 Published on 《 Nature Machine Intelligence》.

GLAM Open source address ： https://github.com/yvquanli/GLAM

Thesis link ： https://www.nature.com/articles/s42256-022-00501-8