Pathway Testing for Longitudinal Omics

A self-contained hypothesis is tested for a given pathway of longitudinal omics. 'SlaPMEG' is a two-step procedure. First, a shared latent process mixed model is fitted over the longitudinal measures of omics in a pathway. This shared model allows deviation from the shared process at subject level (a random intercept, slope, or both per subject) and also at omic level (a random effect per omic). These random effects summarize the longitudinal trend of the observations which can be used to test for group differences using 'Globaltest' in the second step. If the pathway is large or the shared effect is small, the package fits a series of pairwise models and estimates the shared random effects based on them.

SlaPMEG

This package is designed to perform a Shared latent Process Mixed Effects analysis with Globaltest.

Here is a list of functions:

slapmeg fits slapmeg for a single feature-set
multslapmeg fits slapmeg simultaneously for several feature-sets
pairslapmeg fits slapmeg based on a computationally efficient approach
plotslapmeg Plots the estimated random effects within the pathway
print.slapmeg Prints the slapmeg model and results with details
summary.slapmeg Prints the slapmeg model and results
simslapmeg Generates joint lingitudinal observations

Usage

For details explanations and example usage check the help files within package, but here are some tips.

• The data need to be in the conventional genomics format, so columns indicate variables and features, whereas rows indicate subjects and repeated measurements.
• The formula must be supplied as a formula object.
• There is a function to simulate longitudinal observations, so you can give it a try if you do not have a real dataset.
• It is possible to use predefined pathways such as GO, KEGG, Wikipathways, and etc. as long as they are put into a list format, you can take a look at examples of creating such lists in the "Creating the Pathlists" section of rSEA package manual.
• If the feature-set has more than 10 features, the slapmeg function will automatically switch to the pairslapmeg which is computationally more efficient.
• If you had convergance issues with smaller feature-sets, try the pairslapmeg function.
• The plotslapmeg function will give an insight on the source of differential expression.