Using birth-death processes to infer tumor subpopulation structure from live-cell imaging drug screening data

Area of Science:

• Oncology
• Computational Biology
• Systems Biology

Background:

• Tumor heterogeneity, characterized by diverse subpopulations with varying drug responses, presents a major obstacle in cancer therapy.
• Accurate characterization of tumor subpopulations is crucial for developing precise and effective treatment strategies.
• Existing computational frameworks like PhenoPop, while useful, are limited by deterministic models that restrict data analysis.

Purpose of the Study:

• To develop an advanced computational framework for analyzing tumor heterogeneity.
• To overcome the limitations of deterministic models in characterizing drug-response subpopulations.
• To enhance the robustness and information extraction from high-throughput drug screening data.

Main Methods:

• Development of a novel stochastic model based on the linear birth-death process.
• Formulation of dynamic variance to improve data utilization and estimation robustness.
• Adaptation of the model for data exhibiting positive time correlation.

Main Results:

• The proposed stochastic model demonstrates improved model fit and enhanced information extraction compared to deterministic approaches.
• The model effectively formulates dynamic variance, leading to more robust estimations from experimental data.
• Validation on both simulated (in silico) and experimental (in vitro) data confirms the model's advantages.

Conclusions:

• The new stochastic model offers a significant advancement in analyzing tumor heterogeneity from bulk screening data.
• This approach enables more precise characterization of drug-response subpopulations, paving the way for improved cancer therapies.
• The model's adaptability to time-correlated data expands its applicability in biological research.