In the absence of group priors, DAPC infers genetic clusters from sequential K-means and model selections.
The optimal number of clusters (i.e., K, demes) is inferred using sequential K-means and model selection (Jombart et al. 2010).
Using the same methods as mentioned above, pathway expression was summarized using the mean, and models were fit using these pathways as predictors.
In the following, we present FunPat and assess gene selection and clustering performance on a number of simulated datasets with known DE genes and dynamic profiles, in comparison with maSigPro, FPCA, edgeR and the hierarchical, k-means and model-based clustering proposed in [ 12].
First, we fitted an unconditional means and growth model (Table 5).
Unconditional means and growth models, as well as two final conditional models, are presented here.
IEA's IDB Analyzer was utilized for all statistical analyses, including the estimation of percentages, means and regression models.
Models A, B, E, and F are in closer agreement with the arithmetic mean and median models than models C, D, and G.
Average 10 percentile thresholds were applied over all ENMs to gain three binomial models from maximum, mean and minimum models.
Mathematical models B, E, and F show less scatter about the mean and median models but there is no systematically better agreement with the other mathematical models.
