人人添人人澡**人人玩

After three months of fierce competition, the Bristol-Myers Squibb-Molecular Translation Competition organized by Kaggle, a well-known data science competition platform, ended at 8 a.m. Beijing time on June 4th. The competition attracted nearly a thousand data scientists from all over the world. The final winner, called SIMM DDDC, is a joint team composed of Dr. Xiaohong Liu from AlphaMa Biotechnology and two Ph.D. students, Feisheng Zhong and Jiacheng Xiong of the Drug Discovery and Design Center (DDDC) of Shanghai Institute of Materia Medica (SIMM), Chinese Academy of Sciences. The joint team stood out from 874 candidates and won the first prize in this competition. Professors Mingyue Zheng mentored this team with valuable suggestions (Figure 1).

Figure 1. The top-ranked candidates of the molecular translation competition.

This competition aims to extract chemical structures from pictures and convert them into International Chemical Identifier (InChI). In publications such as journals and patents, organic molecules are usually described by chemical structural formulas in images. Automatic identification of chemical structures from such documents allows chemists to obtain information for decision-making quickly. In addition, the InChI text chemical formulas obtained by molecular image translation are text data containing rich semantics. They can be more conveniently integrated with biological, pharmacological, and other data/information described in the same text form to build a coherent database.

Data are the foundation of algorithms. How to effectively obtain high-quality data sets for modeling is an urgent problem in drug research and development. The algorithm developed by the “SIMM DDDC” team can accurately extract the structural information of the compounds from pictures. This method can be used for automatic mining and analyzing real-world chemical literature and patent data, which will significantly promote the construction of biomedical big data, which subsequently will underpin the development of artificial intelligence algorithms in the biomedical field.