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

Area of Science:

• Computational biology
• Cancer research
• Systems biology

Background:

• Tumor heterogeneity, characterized by distinct subpopulations, complicates effective cancer therapy development.
• Accurate characterization of subpopulation structure is crucial for precise treatment strategies.
• Previous computational frameworks like PhenoPop used deterministic models, limiting their analytical power.

Approach:

• Developed a novel stochastic model based on the linear birth-death process to analyze tumor heterogeneity.
• The model incorporates dynamic variance, enhancing data utilization and estimation robustness.
• The stochastic model accommodates experimental data with positive time correlation.

Key Points:

• The stochastic model offers improved model fit and information extraction compared to deterministic approaches.
• It provides more robust estimations by formulating dynamic variance over the experimental horizon.
• The model's adaptability to time-correlated data enhances its applicability.

Conclusions:

• The proposed stochastic model advances the analysis of tumor heterogeneity from drug screening data.
• This approach enables more precise characterization of drug-response subpopulations.
• The model demonstrates advantages in both simulated and experimental data analyses, supporting its utility in cancer research.