Fig. 5From: Pickaxe: a Python library for the prediction of novel metabolic reactionsResults of a similarity sampling expansion of YMDB compounds less than 600Â g/mol using 353 JN1224 rules. A set of 250 KEGG compounds containing between four and nine carbon atoms not in the YMDB database were selected as targets. a The total number of compounds produced per generation. Before each generation, 2000 compounds were sampled to be reacted. b The total number of targets produced per generation and c the similarity distribution of each resultant network. This distribution consists of the maximum similarity value of each network compound to the targets. The vertical line signifies the mean similarity scoreBack to article page