A unified simulation model for understanding the diversity of cancer evolution

This is for a website for presenting results accompnying our manuscript titled “A unified simulation model for understanding the diversity of cancer evolution”. In this work, we examaind system dynamisc of the follwing five models by sensitivity analysis:

The sensitivity analysis is performed by utilizing our newly developed tool termed MASSIVE. By employing a supercomputer at Human Genome Center, the Institute of Medical Science, the University of Tokyo, The MASSIVE first performs a huge number of agent-based simulations with a broad range of parameter settings in a four-dimensional parameter space. For each parameter setting, we performed 50 Monte Carlo trials, from which we obtained averaged values of 13 summary statistics (listed in Table 1) for quantifying simulation results. Single-cell mutation profiles (an example is provided in Figure 1) were also produced from 5 of the 50 Monte Carlo trials. Each single-cell mutation profile is represented as a binary matrix, the row and column indices of which are mutations and samples, respectively. To interpret the simulation results intuitively, we also visualized the binary matrix by utilizing the heatmap function in R. The visualized matrix is accompanied by a left-side blue bar indicating the driver mutations. When the simulated tumor had differentiated cells or subclones with explosive driver mutations, the subpopulation is indicated by the purple bar on the top of the visualized matrix. All the results can be intuitively explored in the focused and view modes of the MASSIVE viewer, which can be visited from the above hyperlinks of the five models. Simulation codes employed for ganarating the data are available from here.

Table 1. a list of the summary statistics

name description
time number of time steps when simulation is finished
population size number of cells when simulation is finished
mutation count per cell mean number of mutations accumulated in each cell
clonal mutation count number of clonal mutations
subclonal mutation count number of subclonal mutations
total mutation count clonal mutation count + subclonal mutation count
clonal mutation proportion clonal mutation count / total mutation count
subclonal mutation proportion subclonal mutation count / total mutation count
Shannon index 0.1 Shannon index calculated with a mutation frequency cutoff of 0.1
Shannon index 0.05 Shannon index calculated with a mutation frequency cutoff of 0.05
Simpson index 0.1 Simpson index calculated with a mutation frequency cutoff of 0.1
Simpson index 0.05 Simpson index calculated with a mutation frequency cutoff of 0.05
driver-branching 0.1 binary statistic indicating that multiple subclones harboring different driver mutations coexist, calculated with a mutation frequency cutoff of 0.1
driver-branching 0.05 binary statistic indicating that multiple subclones harboring different driver mutations coexist, calculated with a mutation frequency cutoff of 0.05
subpopulation proportion proportion of differentiated cells or subclones with explosive driver mutations

Figre 1. an example of the mutation profile heat maps