README.md

In neyhartj/fsimpute: Imputing genotype data in bi-parental populations of finite selfing

fsimpute

An R package for imputing sparse genotyping data in bi-parental populations of partially inbred individuals

Author

Jeff Neyhart

Introduction

Imputation software abounds for low-coverage genotype data generated from platforms such as genotyping-by-sequencing, particularly in the field of plant breeding and genetics. However, there are not many easy-to-use imputation software packages that can handle populations that have undergone finite generations of inbreeding.

fsimpute (full sibling / finite selfing impute) is a simple package for imputation of low-coverage genotype data in bi-parental populations of partially inbred individuals. The software can handle bi-allelic SNP loci. The methodology is based partly on functions implemented in the mpMap package and their imputation procedure for multi-parent populations (Huang et al. 2014).

Simulations

To test the efficacy of fsimpute, I ran a few simulations using fake and real data. The fake data was generated through R/qtl and the real data consisted of genotype data on barley breeding lines. This information was downloaded from the T3 database.

I tested 3 family sizes (30, 100, and 200 individuals), 3 levels of inbreeding (2 selfing generations = $F_3$, 4 selfing generations = $F_5$, and 6 selfing generations = $F_7$), and 4 per-SNP missing marker cutoffs (20%, 40%, 60%, and 80%). 50 replications of each scenario were performed.

Fake Data Simulations

Procedure

You will need three pieces of information to use fsimpute. 1) genotype data on parents, 2) genotype data on the progeny, and 3) a genetic map of the markers.

Installation

The package can be installed via the devtools package:

devtools::install_github("neyhartj/fsimpute")

neyhartj/fsimpute documentation built on May 23, 2019, 4:29 p.m.