The key challenge in plethora of natural sciences – chemistry, biology, biophysics, material science, etc. – is how to understand the behavior of evolving systems consisting of vast number of atoms belonging to different molecules that engage in various interactions. In addition to understanding the fundamental nature of the underlying processes, understanding this behavior is key to applications of societal relevance, such as drug design and discovery, modeling and prediction of protein structures, grain size evolution, etc. Due to safety concerns, as well as (prohibitive) costs, instead of experimental studies – especially in the preliminary research stages – Molecular Dynamics (MD) is an attractive alternative for generating trajectories of atoms that move in a 3D space and are subject to different laws of quantum physics and analytical chemistry. However, analytical expressions cannot be used to characterize every particle, nor can they be used with a confidence for predicting occurrence(s) of events of interest. In the recent years, multiple data analytics methods have been developed to analyze the vast datasets obtained via simulation. In this paper, we address questions related to a possible “push-pull” entaglement between: (1) how and to what extent can spatio-temporal data management and analytics improve the process of knowledge generation in the micro-world of molecular reactions? (2) are there issues from this realm (and using MD data) that can initiate novel challenges for spatio-temporal data management and analytics?