Calculates profile likelihood confidence intervals for results using hapipf
profhap [if exp] , or(string) ipf(string) [ by(string) acc(real) level(real) hapacc(real) savegraph ]
Description
This function calculates the confidence intervals for the odds ratios estimated by the command hapipf. It uses the profile likelihood by estimating the log-likelihood of a series of constrained log-linear models.
This command will only estimate the confidence interval for a single OR and can be used to estimate the odds ratio in a stratified analysis.
Options
or(string) first the case control variable is specified then two haplotypes/alleles. The first haplotype/allele indicates the unexposed group and the other haplotype/allele is the exposed group e.g. or(D 1 2) specifies that the case control variable is D and that the allele 1 represents the unexposed group and allele 2 the exposed.
ipf(string) specifies the log-linear model in the constrained table. It requires special syntax of the form l1*l2+l3. l1*l2 allows all the interactions between the first two loci and locus 3 is independent of them. This syntax is used in most books on Log-linear modelling, "-" terms and brackets are not allowed.
by(string) specifies a stratifying variable.
acc(real) specifies the accuracy of the estimated upper and lower bounds of the confidence interval.
level(real) specifies the significance level of the confidence interval.
hapacc(real) specifies the convergence threshold of both hapipf and ipf.
savegraph specifies that the profile graph is saved to file profile.gph
Examples
Single-point Odds Ratio (OR)
The simplest example is where there is one genetic marker (with alleles 1 and 2) and a binary outcome D. The model specified in the ipf() option must not include the interaction of interest i.e. whether marker l1 is associated with D because this term is fitting using constraint files. Only one OR confidence interval is calculated using profile likelihood and "reference" and "exposed" alleles are specified as 1 and 2, respectively, in the or() option.
. profhap a1 a2, ipf(l1+D) or(D 1 2)
Stratified Estimation of an OR
The parameter of interest is the OR comparing haplotype 2.2 to haplotype 1.1. Where the OR is now the pooled estimate over each level of the categorical variable S. Again the haplotype-outcome interaction l1*l2*D is not included in the ipf() model.
. profhap a1 a2 b1 b2, ipf(S*D+l1*l2*S) or(D 1.1 2.2) by(S) acc(0.0001)
This can be extended to 3 markers (comparing haplotype 2.2.2 to haplotype 1.1.1).
. profhap a1 a2 b1 b2 c1 c2, ipf(S*D+l1*l2*l3*S) or(D 1.1.1 2.2.2) by(S) acc(0. > 0001)
Author
Adrian Mander, Glaxo Smithkline, Harlow, UK.
Email adrian.p.mander@gsk.com
Also see
Related commands
HELP FILES Installation status SSC installation links Descrip > tion
hapipf (required) (ssc install hapipf) Haplotype Frequ > ency estimation qhapipf (if installed) (ssc install qhapipf) Quantitative Tr > ait swblock (if installed) (ssc install swblock) Haplotype block > s hapblock (if installed) (ssc install hapblock) Haploty > pe blocks gipf (if installed) (ssc install gipf) Graphical Log-l > inear Representation