Effectiveness of Grammatical Evolution Decision Trees in Identifying Disease-Causing Polymorphisms

• Activity Number: 188
• Type: Contributed
• Date/Time: Monday, August 1, 2011 : 10:30 AM to 12:20 PM
• Sponsor: ENAR
• Author(s): Kristopher Hoover*+ and Rachael Marceau and Tyndall Harris and David Reif and Alison Motsinger-Reif
• Companies: North Carolina State University and North Carolina State University and North Carolina State University and North Carolina State University and North Carolina State University
• Keywords: evolutionary computation ; statistical genetics ; gene-gene interactions ; decision trees ; variable selection ; genetics

Abstract

The identification of complex genetic models including gene-gene interactions that predict common disease is an important goal of human genetics. Detecting such complex models in high-dimensional data creates an analytical challenge, requiring methods to perform both statistical estimation and variable selection. Recently, a Grammatical Evolution Decision Tree (GEDT) approach has been proposed to detect such complex models. Decision trees are easily interpretable; however their power is limited in identifying strict interactions due to their usual hierarchical model building approach. Grammatical evolution, a type of evolutionary computation, is used to avoid this problem by evolving decision trees to detect interactive models. In the current study, we present the results of parameter sweep optimization for GEDT such as generations, mutation rate, etc. Additionally, we show power comparisons to similar methods such as neural networks and logistic regression approaches.

This material is based upon work supported by the National Science Foundation under the NSF-CSUMS project DMS-0703392 (PI: Sujit Ghosh).