Fig. 1.

Graphical description for Methods a. Entire pipeline to construct our prediction model. b. A brief graphical description about dataset preparation. 1444 molecular descriptors were calculated for substrates in training dataset, and the subtractions of descriptors were calculated for every substrates pair. For supervised learning, a set of descriptor subtractions of substrates pair was labeled with 1 or 0. a, b and c denote substrates in the training dataset. E1 and E2 denotes enzymes in the dataset