Skip to Content

suest {Haplin}

Compute a joint p-value for a list of haplin fits (usually from a sliding window approach), correcting for multiple testing.


The first argument to suest should be a list of haplin estimation results (from the same data file), usually the output from haplinSlide. suest produces as a result a joint overall p-value based on aggregating the individual p-values and then correcting for multiple testing. The correction is achieved by using the principle of "seemingly unrelated" estimation, taking into account the correlation between the individual estimation results.





A list whose elements are different haplin runs on the same data file, typically the output of haplinSlide.


    haplinSlide runs haplin on a series of overlapping windows of markers from the same data file, typically within the same gene. Since each run produces a separate overall p-value, suest computes a joint overall p-value for the gene (or region) that has been scanned. It corrects the overall p-value for multiple testing, also taking into account the fact that the sequence of estimates produced by haplinSlide will be dependent, both because they are computed on the same data set and also since the windows are overlapping (if the window length is larger than 1). If the suest estimation fails (which doesn't happen very often), a standard Bonferroni correction is used instead. Important: haplinSlide must be run with the option table.output = FALSE to provide suest with enough information.


    A list is returned, the most important elements of which are:               

The overall score p-values from each haplin run
The joint p-value, corrected for multiple testing
A logical, usually FALSE, which means the suest estimation went well. If TRUE, it means that the suest estimation failed for some reason, and a standard Bonferroni correction was used instead.


Gjessing HK and Lie RT. Case-parent triads: Estimating single- and double-dose effects of fetal and maternal disease gene haplotypes. Annals of Human Genetics (2006) 70, pp. 382-396.

Web Site:


    Further information is found on the web page.

See Also

haplin, haplinSlide


## Not run:
# (Almost) all standard haplin runs can be done with haplinSlide. 
# Below is an illustration. See the haplin help page for more 
# examples.
# Analyzing the effect of fetal genes, including triads with missing data,
# using a multiplicative response model. When winlength = 1, separate
# markers are used. To make longer windows, winlength can be increased
# correspondingly:
result.1 <- haplinSlide("C:/work/data.dat", use.missing = T, response = "mult",
reference = "", winlength = 1, table.output = F)
# Provide summary of separate results:
lapply(result.1, summary)
# Plot results:
par(ask = T)
lapply(result.1, plot)
# Compute an overall p-value for the scan, corrected for multiple testing
# and dependencies between windows:
## End(Not run)


Hakon K. Gjessing
Professor of Biostatistics
Division of Epidemiology
Norwegian Institute of Public Health

Documentation reproduced from package Haplin, version 5.3. License: GPL (>= 2)