tugHall: a simulator of cancer-cell evolution based on the hallmarks of cancer and tumor-related genes

SMITH: spatially constrained stochastic model for simulation of intra-tumour heterogeneity

MOTIVATION

Simulations of cancer evolution are highly useful to study the effects of selection and mutation rates on cellular fitness. However, most methods are either lattice-based and cannot simulate realistically sized tumours, or they omit spatial constraints and lack the clonal dynamics of real-world tumours.

RESULTS

Stochastic model of intra-tumour heterogeneity (SMITH) is an efficient and explainable model of cancer evolution that combines a branching process with a new confinement mechanism limiting clonal growth based on the size of the individual clones as well as the overall tumour population. We demonstrate how confinement is sufficient to induce the rich clonal dynamics observed in spatial models and cancer samples across tumour types, while allowing for a clear geometric interpretation and simulation of 1 billion cells within a few minutes on a desktop PC.

AVAILABILITY AND IMPLEMENTATION

SMITH is implemented in C# and freely available at https://bitbucket.org/schwarzlab/smith. For visualizations, we provide the accompanying Python package PyFish at https://bitbucket.org/schwarzlab/pyfish.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Dataset of tugHall simulations of cell evolution for colorectal cancer

Dataset contains results of multiple parallel calculations using the tugHall simulator. Output data of simulations are variant allele frequencies for four genes (APC, KRAS, TP53, and PIK3CA) related to colorectal cancer. During each simulation tugHall stochastically reproduces Darwinian evolution for cancer cells and calculates clonal heterogeneity. The probabilities of stochastic processes depend on a correspondence matrix between genome information and cancer hallmarks. As a result, tugHall records variant allele frequencies for the final stage of evolution. The number of trials is several million to get rich statistics of stochastic processes. These data can be used for approximate Bayesian computation and other statistical methods to get personalized coefficients for patients with colorectal cancer. The procedure of usage data is explained in our paper [Bioinformatics, 36, 11 (2020) 3597] in which the part of these data was used.