Summary: are linked, multiple comparisons at the same locus
are correlated, and not all DNA markers are equally
informative. Instead of a Bonferroni correction, we
used an empirical permutation test. We created rep-
licates of the observed sample, in which genotype
and phenotype distributions were preserved, but any
linkage between genotype and phenotype was re-
moved by randomly reassigning the observed pheno-
typic values. We then asked how often any replicates
that conformed to the null hypothesis (the indepen-
dence of genotype and phenotype) produced P values
in excess of the P value observed for the real data.
This test preserved both the observed marker infor-
mativeness and the overall phenotype distribution.
23. E. Lander, L. Kruglyak, Nature Genet. 11, 241 (1995).
24. Linkage of the resistance allele to two distinct marker
alleles, 2 and 4, could be explained by recombination
between resistance and marker alleles or by multiple
entry of the resistance allele into the family. An
attempt to distinguish by reconstructing flanking