A new method to infer causal phenotype networks using QTL and phenotypic information

Area of Science:

• Genetics and Breeding
• Systems Biology
• Bioinformatics

Background:

• Reconstructing causal structures between phenotypic traits is crucial for understanding genetic intervention effects.
• Current methods rely on quantitative trait loci (QTLs), often unmet for molecular phenotypes due to data limitations.
• Existing approaches require identified QTLs for every trait, limiting applicability.

Purpose of the Study:

• To develop a novel algorithm for inferring causal directions in phenotype networks.
• To overcome limitations of existing methods that require QTLs for all traits.
• To improve the accuracy of causal inference in genetic and breeding research.

Main Methods:

• Introduced the QTL+phenotype supervised orientation (QPSO) algorithm, a heuristic search method.
• QPSO does not require a quantitative trait locus (QTL) for every trait studied.
• The algorithm incorporates phenotypic interactions alongside detected QTLs for edge orientation.

Main Results:

• QPSO demonstrates broader applicability compared to the QTL-directed dependency graph (QDG) algorithm.
• The QPSO algorithm achieves more accurate overall orientations of causal relationships.
• Evaluated performance through simulations and a real-life case study with tomato metabolites.

Conclusions:

• QPSO offers a more flexible and accurate approach to inferring causal networks between phenotypic traits.
• The method is particularly valuable for complex traits and molecular phenotypes where QTL identification is challenging.
• QPSO enhances the understanding of genetic architectures and facilitates breeding strategies.