Selection Estimation from Genetic Time-Series Data: Effects of Limited Sampling and Genetic Drift

Area of Science:

• Evolutionary biology
• Population genetics
• Statistical genetics

Background:

• Estimating selection from genetic time-series data is crucial for understanding evolutionary dynamics.
• Accurate selection inference is challenged by noise from limited population sampling and genetic drift.

Purpose of the Study:

• To analyze a selection coefficient estimator under the marginal path likelihood (MPL) framework.
• To characterize how sampling and genetic drift uncertainties affect estimator performance.

Main Methods:

• Mathematical analysis of a selection coefficient estimator derived from the marginal path likelihood (MPL) framework.
• Identification of integrated mutant allele variance as a key determinant of estimator precision.

Main Results:

• Variance integration in MPL mitigates sampling and genetic drift errors at different rates.
• Genetic drift typically emerges as the dominant error source in longer genetic time-series.
• MPL-based estimation demonstrates surprising robustness to sampling errors, despite the underlying model neglecting this factor.

Conclusions:

• Temporal information incorporation effectively reduces multiple noise sources in selection coefficient estimation.
• The integrated mutant allele variance is a critical parameter for assessing estimator precision.
• MPL offers a robust framework for inferring selection from noisy genetic time-series data.