A Bayesian framework for de novo mutation calling in parents-offspring trios

Area of Science:

• Genomics
• Computational Biology
• Human Genetics

Background:

• Spontaneous (de novo) mutations (DNMs) are key factors in complex disease etiology.
• Identifying DNMs in sporadic cases aids in discovering disease-associated genes and genomic regions.
• Next-generation sequencing of parent-proband trios enables genome-wide DNM detection, but is challenged by sequencing noise and artifacts.

Purpose of the Study:

• To develop a novel Bayesian framework for accurate de novo mutation calling in trios.
• To overcome limitations of existing methods by allowing flexible, site-specific mutation rates.
• To improve the sensitivity and specificity of de novo mutation detection.

Main Methods:

• Developed and implemented the TrioDeNovo Bayesian framework in C++.
• Disentangled prior mutation rates from data likelihood for flexible prior adjustment.
• Incorporated filtering based on sequence alignment characteristics to enhance accuracy.

Main Results:

• TrioDeNovo demonstrated improved sensitivity and specificity compared to existing methods via simulations and real data analysis.
• The framework allows for flexible incorporation of priors to further enhance efficiency.
• Filtering based on alignment characteristics significantly improved calling accuracy.

Conclusions:

• TrioDeNovo offers a robust and flexible Bayesian approach for de novo mutation calling in trios.
• The method enhances the accuracy of identifying genetic variants implicated in complex diseases.
• This framework facilitates more reliable genetic studies by improving de novo mutation detection.