GlobalAncova with Special Sum of Squares Decompositions

1 Abstract This vignette shows the enhancements made for GlobalAncova. Basically, there are four ideas implemented: ˆ decomposition of the sum of squares of a linear model ([2]) ˆ a plotting function for the sequential decomposition ˆ pairwise comparison for factor levels ˆ adjustment for global covariates The decomposition of the model sum of squares results in an ANOVA table, which shows the sum of squares due to each term of a linear model. It can be used either on a global basis or for a small group of genes on a gene-wise basis. Pairwise comparisons allow conclusions about whether or not the gene expressions of two levels of a factor are significantly different. An adjustment for global covariates is possible in cases where not only one but two expression sets per subject exist. If the second set describes the subject’s ’normal’ status, it can be used to reduce the variance between subjects.