Integrative modeling of the cell

Abstract

A whole-cell model represents certain aspects of the cell structure and/or function. Due to the high complexity of the cell, an integrative modeling approach is often taken to utilize all available information including experimental data, prior knowledge and prior models. In this review, we summarize an emerging workflow of whole-cell modeling into five steps: (i) gather information; (ii) represent the modeled system into modules; (iii) translate input information into scoring function; (iv) sample the whole-cell model; (v) validate and interpret the model. In particular, we propose the integrative modeling of the cell by combining available (whole-cell) models to maximize the accuracy, precision, and completeness. In addition, we list quantitative predictions of various aspects of cell biology from existing whole-cell models. Moreover, we discuss the remaining challenges and future directions, and highlight the opportunity to establish an integrative spatiotemporal multi-scale whole-cell model based on a community approach.

Keywords: cell biology; compartment model; integrative modeling; whole-cell modeling.

Figure 1

Description of the whole-cell modeling workflow (A) The whole-cell modeling workflow is summarized into five steps using three examples [ 19, 22, 28] . First, gather information such as reactions of components, ODEs and protein distances. Second, represent the modeled system into modules ( e.g., cell compartments, cellular processes and protein communities) based on input information. Third, translate input information into scores, which can be physical restraints (whole-cell compartment model), the sum of square differences (whole-cell mathematical model) and loss functions (whole-cell structure model). Fourth, sample the whole-cell model by integrating intermediate models of modules or information over modules directly, followed by custom optimizations ( e.g., refinement of model parameters and model concordance with previous studies). Fifth, validate the model by information used and not used to construct the model, quantify the model uncertainty, and interpret the model. Iteration of the five steps allows to refine the cell model when there is new information, modeling methods or computational capacity. (B) Integrating available (whole-cell) models will yield a more accurate, precise and complete whole-cell model [24].