Through millions of affinity measurements, researchers have developed a biochemical model that reveals novel insights into microRNAs, which function in part to silence gene expression, but other biological functions of which are unknown. The new model also enables a more accurate prediction of the effects of specific miRNA sequences on genetic expression. Small, non-coding miRNAs regulate most human mRNAs and have been shown to play key roles in several developmental and physiological processes, as well as in disease. miRNAs target and bind with complementary mRNAs and function by silencing sequence-specific gene expression. Although most genes contain targets for at least one miRNA, the diverse biological functions miRNAs enable remain largely unknown. Predicting these functions requires a better understanding of miRNA targeting efficacy and the affinity between a miRNA complex and its target site. Due to the lack of affinity measurements, predictive models of targeting efficacy, like the widely used TargetScan7, generally rely on correlative approaches and can only explain a small fraction of miRNA-specific changes in mRNA. To address this, Sean McGeary and colleagues evaluated the binding affinities between six miRNAs and their artificial targets using RNA Bind-N-Seek, a high-throughput assay which characterizes the sequence and structure of RNA binding proteins. Using these millions of affinity measurements, McGeary et al. developed a biochemical model of miRNA-mediated silencing, which was expanded to include all miRNA sequences using a convoluted neural network (CNN). According to the authors, the biochemical model both explains and predicts nearly half the variability attributable to the direct effects of miRNAs on their specific targets, and greatly outperforms the most powerful correlative models currently being used.