QTDT - Online Documentation

Speeding up analysis of SNP markers

By default, QTDT analyses every allele at every marker locus. For di-allelic markers, both alleles will provide exactly the same evidence for association. When analysing sets consisting solely of biallelic markers (such as SNPs), you may wish to specify that you only want to analyse one allele at each marker. This will reduce the amount of output and may even speed things up a bit:

   prompt> qtdt --snp

Parent-of-origin effects

For total association or transmission disequilibrium tests with the orthogonal model, the effects of maternally and paternally derived alleles may be modelled separately by specifying the -o option:

File	Description
-ot	Tests whether maternal and paternal alleles are significantly different
-of	Uses a saturated model, where paternal and paternally inherited alleles are modelled separately
-om, -op	Test only maternally or paternally derived alleles.

When the imprinting option is requested the informative individuals are two groups of informative individuals. One group is always included in analyses, the other group is only included when testing association at the population level (-at option, which does not control for population stratification):

The first group includes those individuals where both parents are genotyped and where one parent is homozygous or the mother and father have different genotypes. In addition, when paternal imprinting is tested, the father must be heterozygous and, when maternal imprinting is tested, the mother must be heterozygous. When the effects of maternal and paternal alleles are modelled simultaneously, at least one parent must be heterozygous.

The second group includes all individuals with at least one homozygous parent. In this group, genotypes for the other parent may be missing. This group is only included when testing effects of imprinting at the population level.

For example, to test whether there are differences between maternal and paternal alleles:

  prompt> qtdt -ao -ot
 

If you find significant differences you may then wish to consider the saturated model:

   
   prompt> qtdt -ao -of
    

Ridiculously Brief Description of Output

QTDT outputs a short description of the linear and variance models that will be considered for both the null and alternative hypothesis, before running any tests.

For each test run, a brief summary including likelihood/goodness-of-fit statistics, number of informative individuals, and p-value (if <= 0.10) is output.

The regression tables file (regress.tbl by default), includes additional information including parameter estimates under the null and alternative hypothesis.