Mechanistic models of how single cells respond to different perturbations can help integrate disparate big data sets or predict response to varied drug combinations. However, the construction and simulation of such models have proved challenging. Here, we developed a python-based model creation and simulation pipeline that converts a few structured text files into an SBML standard and is high-performance- and cloud-computing ready. We applied this pipeline to our large-scale, mechanistic pan-cancer signaling model (named SPARCED) and demonstrate it by adding an IFNγ pathway submodel. We then investigated whether a putative crosstalk mechanism could be consistent with experimental observations from the LINCS MCF10A Data Cube that IFNγ acts as an anti-proliferative factor. The analyses suggested this observation can be explained by IFNγ-induced SOCS1 sequestering activated EGF receptors. This work forms a foundational recipe for increased mechanistic model-based data integration on a single-cell level, an important building block for clinically-predictive mechanistic models.

The current era of big biomedical data accumulation and availability brings data integration opportunities for leveraging its totality to make new discoveries and/or clinically predictive models. Black-box statistical and machine learning methods are powerful for such integration, but often cannot provide mechanistic reasoning, particularly on the single-cell level. While single-cell mechanistic models clearly enable such reasoning, they are predominantly “small-scale”, and struggle with the scalability and reusability required for meaningful data integration. Here, we present an open-source pipeline for scalable, single-cell mechanistic modeling from simple, annotated input files that can serve as a foundation for mechanistic data integration. As a test case, we convert one of the largest existing single-cell mechanistic models to this format, demonstrating robustness and reproducibility of the approach. We show that the model cell line context can be changed with simple replacement of input file parameter values. We next use this new model to test alternative mechanistic hypotheses for the experimental observations that interferon-gamma (IFNG) inhibits epidermal growth factor (EGF)-induced cell proliferation. Model- based analysis suggested, and experiments support that these observations are better explained by IFNG-induced SOCS1 expression sequestering activated EGF receptors, thereby downregulating AKT activity, as opposed to direct IFNG-induced upregulation of p21 expression. Overall, this new pipeline enables large-scale, single-cell, and mechanistically-transparent modeling as a data integration modality complementary to machine learning.

Large-scale and whole-cell modeling has multiple challenges, including scalable model building and module communication bottlenecks (e.g. between metabolism, gene expression, signaling, etc). We previously developed an open-source, scalable format for a large-scale mechanistic model of proliferation and death signaling dynamics, but communication bottlenecks between gene expression and protein biochemistry modules remained. Here, we developed two solutions to communication bottlenecks that speed up simulation by ~4-fold for hybrid stochastic-deterministic simulations and by over 100-fold for fully deterministic simulations.

Summary Large-scale and whole-cell modeling has multiple challenges, including scalable model building and module communication bottlenecks (e.g. between metabolism, gene expression, signaling, etc). We previously developed an open-source, scalable format for a large-scale mechanistic model of proliferation and death signaling dynamics, but communication bottlenecks between gene expression and protein biochemistry modules remained. Here, we developed two solutions to communication bottlenecks that speed up simulation by ∼4-fold for hybrid stochastic-deterministic simulations and by over 100-fold for fully deterministic simulations. Fully deterministic speed-up facilitates model initialization, parameter estimation and sensitivity analysis tasks. Availability and Implementation Source code is freely available at https://github.com/birtwistlelab/SPARCED/releases/tag/v1.3.0 implemented in python, and supported on Linux, Windows, and MacOS (via Docker).