Live replicator dynamics of the go-or-grow dichotomy in tumor cell populations
Grounded in “Effect of intratumor heterogeneity in managing the go-or-grow dichotomy of cancer cells:
a game theory modeling to understand metastasis” — arXiv:2510.23360.
The paper models tumor cell competition for nutrients and oxygen as a static evolutionary game: each cell
can stay and proliferate in the primary tumor or adopt a motility strategy to seek resources
at a metastatic site. Its evolutionary equilibrium reproduces real oncology observations — the
coexistence of primary and metastatic tumors — and, fit to clear cell renal cell carcinoma (ccRCC)
data, finds that motility is favored in punctuated-evolution cancers specifically when the proportion of
BAP1 mutations stays below a threshold. This demo runs the same class of game — a two-strategy
replicator dynamic — live in your browser, and lets that BAP1-style threshold show up as a heterogeneity
slider.
Poker analogy: a Grow/Go split converging to a fixed ratio where neither strategy can improve its
payoff by deviating is exactly a Nash / evolutionarily stable equilibrium — the same math as a solver's
GTO mixed-strategy frequencies. A tumor cell population "randomizing" 63% stay / 37% go is doing what a
solver does when it bluffs a river 33% of the time: any other ratio gets exploited (by starving neighbors,
or by a bet-into-fold line), so evolution / equilibrium search converges on the one mix that's unbeatable.
Tumor Simulation agent-based, N = 220 cells
Blue = staying & proliferating in the primary tumor. Purple = migrating toward the metastatic site.
Click/tap the canvas to instantly flip a chunk of cells and watch the population fight its way back to
equilibrium — that resistance to invasion is what "evolutionarily stable" means.
Controls & Live Readout
Higher scarcity ⇒ local competition punishes "grow" harder ⇒ equilibrium
shifts toward "go" (cells leave to find resources elsewhere) — directly from the paper's
nutrient/oxygen competition term.
Dashed marker = illustrative BAP1-proportion threshold from the paper's
ccRCC fit. Near the threshold, clonal diversity most favors successful colonization of the metastatic
site (motility payoff peaks); far from it, "grow" wins out. This is a simplified stand-in for the
paper's punctuated-vs-branching threshold effect, not a literal reproduction of its curve.