Autism

Most approaches for rare variants focus mainly on individual gene analysis. However, it has also been recognized that a

majority of biological behaviors manifest from a complex interaction of biological pathway. Although several statistical

methods for pathway-based analyses using rare variants have been proposed to analyze pathways individually, those methods

generally assume that the pathways are independent and do not consider substantial overlapping genes between pathways.

In this paper, we propose a novel statistical approach: a robust pathway-based approach using hierarchical components. The

proposed method uses generalized structural components and analyzes all pathways from the public databases such as Kyoto

Encyclopedia of Genes and Genomes (KEGG) pathway database. To consider the relationship between genes within-pathway,

we use both reflective and formative models. Furthermore, our approach uses ridge-regularized method to reduce the effects

of the overlapping genes between pathways. Evaluation of the method was performed with the simulation data sets as well as

the sequencing data for the phenotype ofautism.