calc.Rsq: Estimate R square of individual QTLs and QTL interactions

Description

Estimates the R square (phenotypic contribution) for individual QTLs and their significant interactions for each trait from cross and peak objects.

Usage

calc.Rsq(cross,peak,th=0.001,round)

Arguments

cross

An object of class cross. See 'qtl' package manual for read.cross function details.

peak

An object of class peak. See define.peak for details.

A numeric vector of length 1 with value between 0 and 1. The R square significance threshold to keep the R square values.

round

An optional integer indicating the precision to be used for the R square value and it significance. The function 'round' is used for R square value rsq. The function 'signif' is used for the significance pF ( i.e. p(F

Value

Return an object of class rsq which is a simple data frame with columns:
qtlThe name of the genetic determinant. If the genetic determinant is an individual QTL, the name is formated as "traitName.chrNumber.orderNumber". In the case of epistasis, the genetic determinant name is formated as the list of individual genetic determinant (QTL) names separated by ":".
rsqThe R square value (set to NA if not significant: pF < th).
pFThe significance (set to NA if not significant: pF < th).

Details

Estimates the proportion of the phenotypic variation explained by the segregation of an individual QTL or significant QTL interactions (also called R square). Compute R square i.e. to compare the phenotypic variation explained by the presence of a genetic determinant (an individual QTL or a QTL's interaction) with the total phenotypic variation. Here we use an AnOVa with a linear model including all possible epistatic interactions.

Let a trait be affected by 3 QTLs localised at 3 markers M1, M2 and M3 respectively. The AnOVa is computed for the linear model: $$M1 + M2 + M3 + M1:M2 + M1:M3 + M2:M3 + M1:M2:M3$$ The R square for each genotypic group is the comparison of the variance due to the between- groups variability (called Mean Square Effect, or MSeffect) with the within- group variability (called Mean Square Error, or Mserror).

The significance of an epistatic interaction is the significance of the group effect of each interaction computed by the AnOVa. If one wants to store the results within a QTL database, it might be useful to compute all genetic determinants by setting th=1 and then to extract the significant results by SQL queries.

Examples

Run this code

data(seed10);

# Genotype probabilities
seed10 <- calc.genoprob( cross=seed10, step=2, off.end=0, error.prob=0,
	 map.function='kosambi', stepwidth='fixed');
seed10 <- sim.geno( cross=seed10, step=2, off.end=0, error.prob=0,
	 map.function='kosambi', stepwidth='fixed');

# Genome scan and QTL detection
out.em <- scanone( seed10, pheno.col=1:50, model='normal', method='hk');
out.peak <- define.peak(out.em, 'all');

# R square computing
out.rsq <- calc.Rsq(seed10,out.peak);

# R square computing without taking account of any significance
out.rsq <- calc.Rsq(seed10,out.peak,th=1);