Related Experiment Videos
Confidence intervals of evolutionary distances between sequences and comparison with usual approaches including the
G Andrieu1, G Caraux, O Gascuel
1Département d'Informatique Fondamentale, L.I.R.M.M., Montpellier, France.
Molecular Biology and Evolution
|August 1, 1997
Summary
Estimating evolutionary distances is crucial for understanding sequence evolution. This study introduces interval estimation as a more reliable method than variance or bootstrap, especially for small evolutionary distances, improving accuracy in phylogenetic analysis.
Area of Science:
- Computational biology
- Bioinformatics
- Evolutionary genetics
Background:
- Estimating evolutionary distances between biological sequences is fundamental in phylogenetics.
- Common methods like variance estimation and bootstrapping can be unreliable, particularly for small evolutionary distances.
- Accurate uncertainty assessment is vital for robust phylogenetic inference.
Purpose of the Study:
- To propose and evaluate interval estimation as a superior statistical method for quantifying uncertainty in evolutionary distance estimates.
- To demonstrate the construction of confidence intervals for Jukes-Cantor and Kimura two-parameter models.
- To compare the accuracy of interval estimation against traditional methods using simulated and real biological data.
Main Methods:
- Development of exact confidence interval construction for Jukes-Cantor and Kimura two-parameter evolutionary distance estimators.
- Comparative analysis using artificial datasets with controlled evolutionary rates and sequence lengths.
- Validation with empirical biological sequence data.
Main Results:
- Traditional methods (variance estimation, bootstrapping) significantly underestimate variability at low substitution rates and short sequence lengths.
- Interval estimation provides more accurate and reliable confidence intervals for evolutionary distances.
- The proposed method highlights limitations of existing approaches in specific evolutionary scenarios.
Conclusions:
- Interval estimation offers a more robust approach to assessing uncertainty in evolutionary distances compared to variance or bootstrap methods.
- The findings underscore the need for improved statistical techniques in phylogenetic analysis, particularly for closely related sequences.
- This work has implications for various evolutionary distance estimators, suggesting a need for re-evaluation of uncertainty quantification in bioinformatics.