Rxivist logo

Partitioning Phenotypic Variance Due To Parent-Of-Origin Effects Using Genomic Relatedness Matrices

By Charles Laurin, Gabriel Cuellar Partida, Gibran Hemani, George Davey Smith, Jian Yang, David M Evans

Posted 31 May 2017
bioRxiv DOI: 10.1101/133827 (published DOI: 10.1007/s10519-017-9880-0)

We propose a method, G-REMLadp, to estimate phenotypic variance attributable to parent-of-origin effects (POEs) across the genome. G-REMLadp uses restricted maximum likelihood analysis of genome-wide genotypic relatedness matrices from phased genotypes. Genome-wide SNP data from parent-child duos/trios are required for relatedness matrices indexing parental origin of offspring alleles. Offspring phenotypes are required to partition trait variation into variance components. We approximated the power of G-REMLadp to detect POEs with Haseman-Elston regression but also used simulations to quantify power, Type I Error rates, and sensitivity to assumptions. We applied G-REMLadp to 38 phenotypes (including body size, obesity, metabolic traits, and IQ) from Avon Longitudinal Study of Parents and Children. Power analyses suggest samples over 10000 unrelated parent-offspring duos are necessary to detect POEs explaining >10% of phenotypic variance. Empirically, it's unlikely that POEs tagged by our genetic relationship matrices explain >15% of variance in any of the 38 traits.

Download data

  • Downloaded 438 times
  • Download rankings, all-time:
    • Site-wide: 109,712
    • In genetics: 4,170
  • Year to date:
    • Site-wide: 198,273
  • Since beginning of last month:
    • Site-wide: 131,048

Altmetric data

Downloads over time

Distribution of downloads per paper, site-wide