Category Archive for : ‘Internal Project Writeups’

A case of a “double banding” pattern with an unusual r² layout on LocusZoom.

Sample duplicate and pedigree mix-up QCs use genetic kinship. There is a base level of relatedness among all HS rats from shared genetic history: outbred from eight inbred strains (Hansen and Spuhler 1984), then maintained as an intrabreeding population. Put blithely, everyone is everyone’s second cousin. Herein two methods to calculate genetic kinship are considered: KING robust coefficients and GRM off-diagonal values. Then, sample duplicates and pedigree discordance QCs for HS rats are developed.

High loadings in principal component analysis and unusual patterns of long-range linkage disequilibrium suggest that Chromosome 13 has a recombination desert, possibly due to an inversion on HS chr13. Large inversions prevent recombination between the inversion and non-inversion haplotypes (Li et al. 2023). We aim to find the location and cause of the recombination desert.

Regional association plots require the positions of genes to plot. We use the refFlat format, as required by LocusZoom. However, the Rat Genome Database (RGD) releases genome annotations in GFF3 format. Thus, we need to convert from one format to the other.

This project seeks to introduce and verify an “albinism QC”. Coat color is easy to phenotype and the genetic basis of albinism in rats is known. An albinism QC compares phenotyped coat color (albino vs. not albino) to genotyped albinism (homozygous recessive vs. others).

Best practice in human GWAS is to include the top principal components (PCs) as covariates. This project evaluated the use of PCs as covariates in GWAS using HS rat data. Does using PC covariates have a significant and/or positive effect on GWAS results?

It is of interest to know the locations and causes of HS rat genotype gaps. For example, little can be done to address a gap due to a lack of founder variation, but a gap due to poor genotyping may be amenable to improved genotyping methods. This project set out to discover gap locations in the context of minor allele frequency (MAF), both in founders and the modern population.

Ties are a problem for rank-based normalization. All samples must be assigned distinct ranks. Therefore, all ties must be broken in some way. Tie-breaking methods include the order samples appear, the alphanumerical order of sample identifiers, or randomly by a pre-determined seed number. We seek to determine the extent to which particular tie-break choices affect the GWAS results.