perm1

genehunter man page (version 1.1, dec 1996)

PERM1

Summary: permutation test for determining TDT significance
Argument: < number of simulations >

Since the standard application of the TDT usually involves the analysis of numerous alleles at numerous markers, a significant correction is required to interpret the significance of any one result. Treating the num_markers x num_alleles tests as independent is extremely conservative since a) the tests of the alleles at each marker are not independent and b) there will be very rare alleles which will penalize an additional degree of freedom without any chance of providing results of interest. A better test of significance is provided by a permutation method in the 'perm1' command as follows:

create a new data set by taking each pair of transmitted and untransmitted alleles and arbitrarily (at p=0.50) reversing the assignment of which was transmitted
tally and store the results of the TDT for this new data set
repeat 1000 or more (the number of simulations is indicated in the argument to 'perm1'), comparing each simulated data set to the actual results observed in the real data set

After the simulations are completed, a report indicating

how many of the permuted data sets had a higher maximum value and
how many of the permuted data sets had more results above certain thresholds (.01, .001)

is displayed providing a better estimate of the significance of the observed data. This command is only available after the 'tdt' analysis of a pedigree file.

PERMUTATION SUMMARY:
12 of 1000 simulations had a larger maximum value than the real best (15.42)
48 of 1000 simulations had as many tests (22) exceeding p=.01
19 of 1000 simulations had as many tests (3) exceeding p=.001