calc.penalties: Calculate LOD penalties

Description

Derive penalties for the penalized LOD scores (used by stepwiseqtl) on the basis of permutation results from a two-dimensional, two-QTL scan (obtained by scantwo).

Usage

calc.penalties(perms, alpha=0.05, lodcolumn)

Value

Vector of three values indicating the penalty on main effects and heavy and light penalties on interactions, or a matrix of such results, with each row corresponding to a different phenotype.

If the input permutations are X-chromosome-specific, the result has six values: main effect for autosomes, main effect for X chr, heavy penalty on A:A interactions, light penalty on A:A interactions, penalty on A:X interactions, and penalty on X:X interactions.

Arguments

perms: Permutation results from scantwo.
alpha: Significance level.
lodcolumn: If the scantwo permutation results contain LOD scores for multiple phenotypes, this argument indicates which to use in the summary. This may be a vector. If missing, penalties for all phenotypes are calculated.

Author

Karl W Broman, broman@wisc.edu

Details

Thresholds derived from scantwo permutations (that is, for a two-dimensional, two-QTL genome scan) are used to calculate penalties on main effects and interactions.

The main effect penalty is the 1-alpha quantile of the null distribution of the genome-wide maximum LOD score from a single-QTL genome scan (as with scanone).

The "heavy" interaction penalty is the 1-alpha quantile of the null distribution of the maximum interaction LOD score (that is, the $\log_{10}$ likelihood ratio comparing the best model with two interacting QTL to the best model with two additive QTL) from a two-dimensional, two-QTL genome scan (as with scantwo).

The "light" interaction penality is the difference between the "fv1" threshold from the scantwo permutations (that is, the 1-alpha quantile of the LOD score comparing the best model with two interacting QTL to the best single-QTL model) and the main effect penalty.

If the permutations results were obtained with perm.Xsp=TRUE, to give X-chr-specific results, six penalties are calculated: main effect for autosomes, main effect for X chr, heavy penalty on A:A interactions, light penalty on A:A interactions, penalty on A:X interactions, and penalty on X:X interactions.

References

Manichaikul, A., Moon, J. Y., Sen, Ś, Yandell, B. S. and Broman, K. W. (2009) A model selection approach for the identification of quantitative trait loci in experimental crosses, allowing epistasis. Genetics, 181, 1077--1086.

Examples

Run this code

data(fake.f2)
fake.f2 <- subset(fake.f2, chr=18:19)
fake.f2 <- calc.genoprob(fake.f2, step=5)
out.2dim <- scantwo(fake.f2, method="hk")

# permutations
permo.2dim <- scantwo(fake.f2, method="hk", n.perm=2)
if (FALSE) permo.2dim <- scantwo(fake.f2, method="hk", n.perm=1000)
summary(permo.2dim, alpha=0.05)

# penalties
calc.penalties(permo.2dim)

Run the code above in your browser using DataLab

Last chance! 50% off unlimited learning