Identifying the factors that determine microbial growth rate under various environmental and genetic conditions is a major challenge of systems biology. While current genome-scale metabolic modeling approaches enable us to successfully predict a variety of metabolic phenotypes, including maximal biomass yield, the prediction of actual growth rate is a long standing goal. This gap stems from strictly relying on data regarding reaction stoichiometry and directionality, without accounting for enzyme kinetic considerations. Here we present a novel metabolic network-based approach, MetabOlic Modeling with ENzyme kineTics (MOMENT), which predicts metabolic flux rate and growth rate by utilizing prior data on enzyme turnover rates and enzyme molecular weights, without requiring measurements of nutrient uptake rates. The method is based on an identified design principle of metabolism in which enzymes catalyzing high flux reactions across different media tend to be more efficient in terms of having higher turnover numbers. Extending upon previous attempts to utilize kinetic data in genome-scale metabolic modeling, our approach takes into account the requirement for specific enzyme concentrations for catalyzing predicted metabolic flux rates, considering isozymes, protein complexes, and multi-functional enzymes. MOMENT is shown to significantly improve the prediction accuracy of various metabolic phenotypes in E. coli, including intracellular flux rates and changes in gene expression levels under different growth rates. Most importantly, MOMENT is shown to predict growth rates of E. coli under a diverse set of media that are correlated with experimental measurements, markedly improving upon existing state-of-the art stoichiometric modeling approaches. These results support the view that a physiological bound on cellular enzyme concentrations is a key factor that determines microbial growth rate.

One of the most challenging recommendation tasks is recommending to a new, previously unseen user. This is known as the user cold start problem. Assuming certain features or attributes of users are known, one approach for handling new users is to initially model them based on their features.Motivated by an ad targeting application, this paper describes an extreme online recommendation setting where the cold start problem is perpetual. Every user is encountered by the system just once, receives a recommendation, and either consumes or ignores it, registering a binary reward.We introduce One-pass Factorization of Feature Sets, OFF-Set, a novel recommendation algorithm based on Latent Factor analysis, which models users by mapping their features to a latent space. OFF-Set is able to model non-linear interactions between pairs of features, and updates its model per each recommendation-reward observation in a pure online fashion. We evaluate OFF-Set against several state of the art baselines, and demonstrate its superiority on real ad-targeting data.

One of the most challenging recommendation tasks is recommending to a new, previously unseen user. This is known as the user cold start problem. Assuming certain features or attributes of users are known, one approach for handling new users is to initially model them based on their features.Motivated by an ad targeting application, this paper describes an extreme online recommendation setting where the cold start problem is perpetual. Every user is encountered by the system just once, receives a recommendation, and either consumes or ignores it, registering a binary reward.We introduce One-pass Factorization of Feature Sets, OFF-Set, a novel recommendation algorithm based on Latent Factor analysis, which models users by mapping their features to a latent space. Furthermore, OFF-Set is able to model non-linear interactions between pairs of features. OFF-Set is designed for purely online recommendation, performing lightweight updates of its model per each recommendationreward observation. We evaluate OFF-Set against several state of the art baselines, and demonstrate its superiority on real ad-targeting data.