Convex combination sequence kernel association test for rare-variant studies | JoVE Visualize

Area of Science:

Genetics
Statistical Genetics
Genomic Association Studies

Background:

Rare variants play a significant role in complex diseases.
Existing methods for rare-variant association studies often struggle with power and accurate weighting of single-nucleotide variants (SNVs).
Leveraging multiple SNV annotations can potentially improve the detection of associations.

Purpose of the Study:

To propose a novel variant set test that integrates multiple SNV annotations for rare-variant association studies.
To optimize the combination of different sequence kernel association test (SKAT) statistics using multiple kernel learning.
To evaluate the performance of the proposed method in terms of type I error control and statistical power through simulations and real data analysis.

Main Methods:

Developed a novel variant set test by optimizing a convex combination of SKAT statistics, each derived from different SNV annotations.
Employed a multiple kernel learning algorithm to determine optimal combination weights for the SKAT statistics.
Validated the method using data splitting for empirical evaluation and simulations to assess type I error and power.
Applied the method to the Framingham Heart Study (FHS) dataset to identify SNV sets associated with fasting glucose levels.

Main Results:

The proposed method demonstrated preserved type I error rates at a stringent alpha level ( $α = 2.5 \times 1 0^{- 6}$ ).
In simulations with large sample sizes ( $N \geq 5, 000$ ), the method showed greater power than SKAT(-O) when SNV weights were correctly specified.
In the FHS cohort, no genome-wide significant associations were found for fasting glucose with rare variants in 4-kb windows ( $p < 1 0^{- 7}$ ).
Suggestive associations were identified between fasting glucose and rare variants near ROCK2 ( $p = 2.1 \times 1 0^{- 5}$ ) and within CPLX1 ( $p = 5.3 \times 1 0^{- 5}$ ).

Conclusions:

The novel variant set test effectively integrates multiple SNV annotations for enhanced power in rare-variant association studies.
The identified suggestive associations with ROCK2 and CPLX1 warrant further investigation and replication in independent cohorts.
Functional genomic studies are necessary to validate the biological relevance of these findings for glucose metabolism and related pathways.