kTWAS: integrating kernel machine with transcriptome-wide association studies improves statistical power and reveals...

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Genotype-phenotype association studies benefit from aggregating multiple genetic variants.
• Current methods like Transcriptome-wide Association Studies (TWAS) and Sequence Kernel Association Test (SKAT) offer complementary approaches (feature selection and aggregation, respectively).
• No existing methods integrate the strengths of both TWAS and kernel-based approaches.

Purpose of the Study:

• To develop a novel method, kernel-based TWAS (kTWAS), that combines feature selection from TWAS with the aggregation power of kernel methods.
• To compare the performance of kTWAS against existing TWAS and SKAT protocols.
• To identify novel disease-associated genes using real-world genetic data.

Main Methods:

• Developed kTWAS, integrating TWAS-like feature selection into a SKAT-like kernel association test.
• Conducted extensive simulations to evaluate kTWAS performance.
• Applied kTWAS to Wellcome Trust Case Control Consortium and MSSNG (Autism) datasets.

Main Results:

• kTWAS demonstrated higher statistical power compared to TWAS and various SKAT-based methods in simulations.
• Novel disease-associated genes were identified in both Wellcome Trust Case Control Consortium and MSSNG datasets.
• The source code for kTWAS and simulation data are publicly available.

Conclusions:

• kTWAS effectively combines the advantages of TWAS and kernel methods for enhanced genotype-phenotype association mapping.
• The novel kTWAS method offers improved power for genetic studies and aids in discovering disease-associated genes.
• This approach advances the field of statistical genetics and provides a valuable tool for researchers.