BESTish: A Diffusion-Approximation Framework for Inferring Selection and Mutation in Clonal Hematopoiesis.

Area of Science:

• Hematology
• Computational Biology
• Genetics

Background:

• Clonal hematopoiesis (CH) is driven by somatic mutations in hematopoietic stem cells (HSCs).
• Understanding CH evolution requires accurate modeling of variant allele frequency (VAF) dynamics.
• Environmental factors can influence HSC fitness and clonal expansion.

Purpose of the Study:

• To develop a novel stochastic model for analyzing CH trajectories.
• To create an efficient and accurate Bayesian inference method (BESTish) for CH driver analysis.
• To quantify mutation fitness, rate, and environmental effects in CH.

Main Methods:

• Developed a state-dependent stochastic model of wild-type and mutant HSCs.
• Derived mean-field dynamics and Gaussian-Markov approximation for VAF.
• Introduced BESTish, a Bayesian inference method using closed-form distributions for VAF data.

Main Results:

• Mean VAF trajectory follows a logistic form influenced by selective advantage.
• Environmental factors impact VAF variance and autocovariance, not the mean trajectory.
• BESTish provides consistent mutation fitness and rate estimates across studies.
• Identified patient-specific heterogeneity in mutation selection and non-homeostatic environments.

Conclusions:

• BESTish offers a unified framework for quantifying CH evolution.
• The method accurately estimates key parameters like mutation fitness and environmental strength.
• Reveals patient-specific clonal dynamics and identifies mutations in growth-facilitating environments.