Statistical genetics of aging-related genomic and phenotypic change

National Institute On Aging

Investigators

Linked publications & trials

Abstract

To help to analyze and understand aging-related "complex" traits that are affected by many genes and environmental factors, we have followed the path of developing statistical algorithms for the analyses of high-throughput sequencing studies. Our proposed new computational tools provide means to analyze additional types of data (e.g., to identify mitochondrial DNA (mtDNA) variants and to estimate mtDNA copy number efficiently from whole-genome sequences). For experimental tests of the algorithms, we are capitalizing on the special advantages of the BLSA (Baltimore Longitudinal Study of Aging, see Annual Report AG000775), InCHIANTI (see Annual Report AG001050), SardiNIA (see Annual Report AG000675), and UK Biobank projects to help in the assembly of mitochondrial sequence data and multiple phenotypic data in the four cohorts. In order to conduct analyses on large-scale consortium data to study mtDNA variation and copy number, we have developed two computational programs, providing a general solution for the analysis of mtDNA dynamics based on whole-genome sequencing studies. One program (mitoCaller) is designed specifically to identify mtDNA variants; the other (mitoCalc) infers mtDNA copy number in a cell directly from genome sequences. Applying the programs to leukocyte sequences of 2,000 SardiNIA participants and 1,000 InCHIANTI participants, we have shown that heteroplasmies (mtDNA variants with more than one allele at a site) increase with age, and that copy number is relatively highly heritable and is correlated with metabolic traits, particularly central fat levels. In more recent work, we have increased the speed of mitoCalc 100-fold (fastMitoCalc). The new program is being applied to white cells of 65,000 deeply sequenced individuals (TOPMed program, NHLBI), for GWAS on copy number. We are also applying our programs to the 500,000 whole-genome sequences from the UK Biobank project. Expanding our other focus on the mtDNA copy number (mtDNAcn) analysis, we have been examining the association between mtDNAcn and personality in participants of the Baltimore Longitudinal Study of Aging (BLSA). We assess the big five personality traits and facets using the Revised NEO Personality Inventory (NEO-PI-R) and estimate mtDNAcn efficiently from whole-genome DNA sequences. Our preliminary analyses show that mtDNAcn is significantly associated with specific domains of the personality inventory. We have also performed mediation analysis to show that mtDNAcn mediates the association between personality and mortality risk. To our knowledge, this is the first study to show a replicable association between mtDNAcn and personality. The results support our hypothesis that mtDNAcn is a biomarker of the biological process that explains part of the association between personality and mortality. We are currently replicating the leading results using data from the UK Biobank project. In a separate study, we are advancing our methodological work by developing deep neural network models to predict phenotypes (e.g., height, Alzheimer’s disease status) directly from raw genomic data. Our goal is to establish a functional pipeline that leverages diverse neural network architectures to analyze single-nucleotide variation, integrating both nuclear SNPs and mitochondrial DNA variants, for phenotype prediction.