Meta-Dimensional Analysis of Phenotypes Using the Analysis Tool for Heritable and Environmental Network Associations (ATHENA): Challenges with Building Large Networks

Abstract

The search for the underlying heritability of complextraits has led to an explosion of data generation andanalysis in the field of human genomics. With thesetechnological advances, we have made some progress inthe identification of genes and proteins associatedwith common, complex human diseases. Still, ourunderstanding of the genetic architecture of complextraits remains limited and additional research isneeded to illuminate the genetic and environmentalfactors important for the disease process, much ofwhich will include looking at variation in DNA, RNA,protein, etc. in a meta-dimensional analysis framework.We have developed a machine learning technique, ATHENA:Analysis Tool for Heritable and Environmental NetworkAssociations, to address this issue of integrating datafrom multiple '-omics' technologies to identify modelsthat explain or predict the genetic architecture ofcomplex traits. In this chapter, we discuss thechallenges in handling meta-dimensional data usinggrammatical evolution neural networks (GENN) which areone modelling component of ATHENA, and acharacterisation of the models identified in simulationstudies to explore the ability of GENN to buildcomplex, meta-dimensional models. Challenges remain tofurther understand the evolutionary process for GENN,and an explanation of the simplicity of the models.This work highlights potential areas for extension andimprovement of the GENN approach within ATHENA.