University of Michigan Center for Statistical 


GOLD - ldmax documentation


ldmax uses the expectation-maximization algorithm of Slatkin and Excoffier (1995), Mol Biol Evol 12:921-7 to estimate haplotype frequencies.


A linkage format pedigree file (pre-makeped) and a map file with marker names and locations. These are specified by the -p and -m parameters respectively.

Unless the -f option is used, only founder genotypes are considered. Rare alleles with frequencies of less than 10% will be pooled into a single allele (this threshold can be changed with the -t option).

For example to analyze the genotypes for the markers described in map.dat and stored in the file ped.dat, run the command:

> ldmax -pped.dat -mmap.dat

To specify that alleles rarer than 5% should be pooled, run:

> ldmax -pped.dat -mmap.dat -t0.05


Haplotype frequencies for all marker pairs are calculated (these can be output to the screen if the verbose -v option is used). A table summarising linkage disequilibrium statistics in the region is output to the screen and to a file named LD.XT. This table is compatible with the graphical interface.


The allele pooling code can cause alleles to be renumbered, so that allele numbers in haplotype frequency estimates do not correspond to original alleles in the pedigree file.


University of Michigan | School of Public Health | Abecasis Lab