Gene Association Analysis of Quantitative Trait Based on Functional Linear Regression Model with Local Sparse...

Area of Science:

• Genetics and Genomics
• Statistical Bioinformatics

Background:

• Functional linear regression models are powerful for gene association analysis of complex traits, utilizing spatial information in genetic variation data.
• Existing high-power methods can produce false associations due to noise, misidentifying non-causal SNPs as significant.

Purpose of the Study:

• To develop a novel method, the Sparse Functional Data Association Test (SFDAT), for gene region association analysis.
• To enhance the accuracy of gene association studies by reducing false positives and improving the handling of noise.

Main Methods:

• Developed SFDAT based on a functional linear regression model incorporating local sparse estimation.
• Introduced novel evaluation indicators, CSR and DL, to assess method performance.
• Compared SFDAT with existing methods like OLS and Smooth using simulation studies.

Main Results:

• SFDAT demonstrated robust performance across linkage equilibrium and disequilibrium simulations for diverse variant types (common, low-frequency, rare, mixed).
• SFDAT showed comparable power and type I error rates to OLS and Smooth, with superior handling of zero regions.
• Analysis of the *Oryza sativa* dataset confirmed SFDAT's ability to improve gene association analysis and eliminate false positives.

Conclusions:

• SFDAT effectively reduces noise interference while maintaining high statistical power in gene association studies.
• The proposed SFDAT method offers a significant advancement for analyzing associations between gene regions and quantitative traits.