2015 IEEE CONFERENCE ON VISUAL ANALYTICS SCIENCE AND TECHNOLOGY(2015)
Christian Michelsen Res
被引用1|浏览6
摘要
De novo design is a computational-chemistry method, where a computer program utilizes an optimization method, in our case an evolutionary algorithm, to design compounds with desired chemical properties. The optimization is performed with respect to a quantity called fitness, defined by the chemists. We present a tool that connects interactive visual analysis and evolutionary algorithm-based molecular design. We employ linked views to communicate different aspects of the data: the statistical distribution of molecule fitness, connections between individual molecules during the evolution and 3D molecular structure. The application is already used by chemists to explore and analyze the results of their evolution experiments and has proved to be highly useful.