The principles of whole-cell modeling Jonathan R Karr 1 , Koichi Takahashi 2,3 and Akira Funahashi 4 Whole-cell models which comprehensively predict how phenotypes emerge from genotype promise to enable rational bioengineering and precision medicine. Here, we outline the key principles of whole-cell modeling which have emerged from our work developing bacterial whole-cell models: single- cellularity; functional, genetic, molecular, and temporal completeness; biophysical realism including temporal dynamics and stochastic variation; species-specificity; and model integration and reproducibility. We also outline the whole-cell model construction process, highlighting existing resources. Numerous challenges remain to achieving fully complete models including developing new experimental tools to more completely characterize cells and developing a strong theoretical understanding of hybrid mathematics. Solving these challenges requires collaboration among computational and experimental biologists, biophysicists, biochemists, applied mathematicians, computer scientists, and software engineers. Addresses 1 Department of Genetics & Genomic Sciences, Icahn School of Medicine at Mount Sinai, New York, NY 10029, USA 2 RIKEN Quantitative Biology Center, RIKEN, Osaka 565-0874, Japan 3 Institute for Advanced Biosciences, Keio University, Fujisawa 252- 8520, Japan 4 Department of Biosciences and Informatics, Keio University, Yokohama 223-8522, Japan Corresponding author: Karr, Jonathan R (karr@mssm.edu) Current Opinion in Microbiology 2015, 27:18–24 This review comes from a themed issue on Microbial systems biology Edited by Eric Brown and Nassos Typas http://dx.doi.org/10.1016/j.mib.2015.06.004 1369-5274/# 2015 Elsevier Ltd. All rights reserved. Introduction Whole-cell models are computational models which de- scribe how phenotype arises from genotype [1,2,3  ]. The primary goal of whole-cell modeling is to enable rational bioengineering and precision medicine. Combined with genome synthesis [4] and transplantation [5], whole-cell models could enable bioengineers to maximize objectives such as biofuel production by optimizing genomes [6,7]. Such models could also enable clinicians to individualize therapy [8–10]. Furthermore, whole-cell models could be powerful scientific tools. Shuler et al. introduced the first coarse-grained ordinary differential equation whole-cell model in 1979 [11,12  ]. Twenty years later, when sequencing provided the first biological parts list, Tomita et al. [13  ] developed the first large-scale fine-grained dynamical model. Researchers have continued to develop increasingly sophisticated dynamical models [14–16]. In parallel, Varma and Palsson used flux balance analysis (FBA) to create the first static genome-scale metabolic models [17]. The latest FBA models represent over 1000 genes [18]. Researchers have since expanded FBA to represent transcriptional regula- tion [19], transcription and translation [20  ], and signaling [21]. Logical methods have also been used [22]. Recently, we and others used a hybrid methodology to construct the first dynamical model which represents every known molecular species and gene function [23  ,24,25]. Simul- taneously, Roberts et al. developed the first cell-scale structural model [26  ]. Here, we describe the core principles of whole-cell modeling. We also outline our model construction pro- cess, highlighting existing tools and the challenges to achieving complete models. The principles of whole-cell modeling Building on Roberts’ discussion [27], we outline 11 fun- damental and practical principles of whole-cell modeling to illuminate a path toward complete models (Figure 1). Single-cellularity First, whole-cell models should represent individual cells. Single-cell models can account for how temporal dynam- ics and stochastic variation affect behavior. Single cells are also tractable because they are independent and directly result from molecular biochemistry. Further- more, single-cell models can take advantage of the grow- ing wealth of single-cell data. Functional closure Behavior is determined by interacting pathways and genes. Consequently, whole-cell models should represent every known cellular and gene function. Models which represent every known function are powerful tools. For example, genome-scale metabolic models which repre- sent every known metabolic reaction and enzyme have been used to identify missing reactions and enzymes [28]. Molecular closure Whole-cell models should represent the cell and its envi- ronment as a closed system. Models should explicitly account for exchanges among pathways and the environ- ment and not have arbitrary sources and sinks. This ensures Available online at www.sciencedirect.com ScienceDirect Current Opinion in Microbiology 2015, 27:18–24 www.sciencedirect.com