A framework for
ecological cascade therapy

in cancer

György Barabás, Gaurav Baruah, Yun Tao, Yie Deng, Yi Dong, Anugraha Puthenpura & Anuraag Bukkuri

2026-07-22

Immune evasion in colorectal carcinoma

The feedback between ecology and evolution

EcoEvoSim automates this workflow

Defining a simulation

using EcoEvoSim

config = EcoEvoConfig(
    ecoDyn = unstructuredModel() do i, n, z, aux, S
       n[i] * (1 - z[i][1]^2 - sum(n[j] for j in 1:S))
    end,
    mutationGenerator = generateMutant(
        invaderPopsize = 0.001,
        variance = 0.002^2
    ),
    integrationParams = IntegrationParams(maxTime = 1e10),
    extThreshold = 0.003
)

initCommunity = Community([1.0], [0.2])
evoHistory = evolve(initCommunity, config, 1500)

The competition-proliferation tradeoff



The competition-proliferation tradeoff

The competition-proliferation tradeoff

\[ \frac{\text{d}}{\text{d} t} n(z_i) = n(z_i) \left[ b(z_i) - \sum_{j=1}^S a(z_i - z_j) n(z_j) \right] \]


b(z) = (tanh(sum(z .- 0.5) / 0.2) + 1)/2 - (tanh(-0.5 / 0.2) + 1)/2
a(z) = (tanh(sum(z) / 0.12) + 1)/2

ecology = unstructuredModel() do i, n, z, aux, S
    n[i] * ( b(z[i]) - sum(a(z[i] .- z[j]) * n[j] for j in 1:S) )
end

Cancer cell diversification

Ecological cascade therapy

Largest tumor density reduction?   ⇒   inverse matrix

Direction of largest tumor density reduction

\[ \frac{\text{d}}{\text{d} t} n(z_i) = n(z_i) \left[ b(z_i) - \sum_{j=1}^S a(z_i - z_j) n(z_j) \right] \]

Direction of largest tumor density reduction

\[ \frac{\text{d}}{\text{d} t} n_i = n_i \left[ b_i - \sum_{j=1}^S a_{ij} n_j \right] \]

\(n_i^* = \sum_j (a^{-1})_{ij} b_j \;\implies\; \Delta n_i^* = \sum_j (a^{-1})_{ij} \Delta b_j\)

\(\min_{\Delta\mathbf{b}}\left(\sum_i \Delta n_i^* \right) = \min_{\Delta\mathbf{b}}\left(\sum_i \sum_j (a^{-1})_{ij} \Delta b_j \right)\)

⇒   Cauchy–Schwarz: \(\;\displaystyle \Delta b_i \sim -\sum_j (a^{-1})_{ji}\)

Perturbing the growth rates

Conclusions

  • Tumors are ecosystems
  • We may understand their structure via eco-evo theory
  • Could help in devising new classes of therapies

Acknowledgments

  • Anuraag Bukkuri
  • Emily Dolson
  • Jie Deng
  • Yun Tao
  • Gaurav Baruah
  • Yi Dong
  • Anugraha Puthenpura