A Family-Informed Phenotype Imputation Approach for Genetic Analyzes

• Activity Number: 355 - Analysis of Complex Genetic Data
• Type: Contributed
• Date/Time: Tuesday, July 31, 2018 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics in Genomics and Genetics
• Author(s): Yuning Chen and Gina Marie Peloso and Ching-Ti Liu and Anita L. DeStefano and James B. Meigs and Josee Dupuis*
• Companies: Boston University and Boston University and Boston University and Boston University and Massachusetts General Hospital, Harvard Medical School and Boston University School of Public Health
• Keywords: Imputation; Missing data; Genome-wide association studies

Abstract

Statistical power is a limitation of genome-wide association studies (GWAS). Sample size is a major component of statistical power that can be easily affected by missingness in phenotypic data and restrain the ability to detect associated single nucleotide variants (SNVs) with small effect sizes. While some phenotypes are hard to collect due to cost and loss of follow-up, correlated phenotypes that are easily collected and complete can be leveraged. We develop a phenotype imputation method incorporating family structure and correlation between multiple phenotypes. We derive the approximated non-centrality parameter (NCP) of the test statistic for association using imputed phenotype values and propose a new approach to analyze the imputed and observed phenotype values in GWAS. We investigate the performance of our method under different conditions and identify situations where our method can increase power in GWAS. We show that our method has a higher imputation accuracy than two other existing methods in family data and investigate factors affecting the imputation accuracy. Finally, we apply the method to fasting glucose and 2 hour glucose data in the Framingham Heart Study.