project.eigen: Infers eigengenes for given expression data
In Pigengene: Infers biological signatures from gene expression data

Description Usage Arguments Details Value Note See Also Examples

This function projects (new) expression data onto the eigengenes of modules from another dataset. It is usfull for comparing the expression behaviour of modules accross (biologically related yet independent) datasets, for evaluating the performance of a classifier on new datasets, and for examining the robustness of a pattern with regards to missing genes.

1 2	project.eigen(Data, saveFile = NULL, pigengene, naTolerance = 0.05, verbose = 0, ignoreModules = c())

`Data`	A matrix or data frame of expression data to be projected. Genes correspond to columns, and rows correspond to samples. Rows and columns must be named. It is OK to miss a few genes originally used to compute the eigengenes, thereby, projection is robust to choose of platform.
`saveFile`	If not `NULL`, where to save the results in `.RData` format.
`pigengene`	An object of `pigengene-class`, usually created by `compute.pigengene`
`naTolerance`	Upper threshold on the fraction of entries per gene that can be missing. Genes with a larger fraction of missing entries are ignored. For genes with smaller fraction of NA entries, the missing values are imputed from their average expression in the other samples. See `check.pigengene.input`.
`verbose`	The integer level of verbosity. 0 means silent and higher values produce more details of computation.
`ignoreModules`	A vector of integers. In order to speed up the projection, it may be desirable to focus only on the eigengenes of a few interesting modules. In that case, the remaining modules can be listed here and will be ignored during projection (Optional).

For each module, from the pigengene object, the weight (membership) of each gene is retrieved. The eigengene is computed (inferred) on the new data as alinear combination using the corresponding weights. The inferred eigengene vector will be normalized so that it has the same Euclidean norm as the original eigengene vector.

A list of:

`projected`	The matrix of inferred (projected) eigengenes
`replacedNaNum`	The number of `NA` entries in the input `Data` that were replaced with the the average expression of the corresponding gene
`tooNaGenes`	A character vector of genes that were ignored because they had too many `NA`s
`notMatched`	A character vector of genes in the original eigengene that could not be matched in the given input `Data`

The new data should use the same type of biolocal identifiers (e.g. Gene Symbols or ENTREZIDs) as the original data for which the pigengene was constructed. It is, however, not required that the new data originate from the same type of technology, e.g. the eigengenes can be based on microarray experiments, whereas the new data comes from an RNA-Seq experiment. Nor is it necessary that the new datset contains measurements for all of the genes from the original modules.

Pigengene-package, compute.pigengene moduleEigengenes

     ## Data:
     data(aml)
     data(mds)
     data(eigengenes33)
     d1 <- rbind(aml,mds)
     Labels <- c(rep("AML",nrow(aml)),rep("MDS",nrow(mds)))
     names(Labels) <- rownames(d1)
     toyModules <- eigengenes33$modules[colnames(d1)]
     ## Computing:
     p1 <- compute.pigengene(Data=d1, Labels=Labels, modules=toyModules, 
        saveFile="pigengene.RData", doPlot=TRUE, verbose=3)
     ## How robust projecting is?
     p2 <- project.eigen(Data=d1, pigengene = p1, verbose = 1)
     plot(p1$eigengenes[,"ME1"],p2$projected[,"ME1"])
     cor(p1$eigengenes[,"ME1"],p2$projected[,"ME1"])