stam.evaluate {stam}R Documentation

StAM Evaluation Procedure


This performs a structured analysis of microarrays (StAM) from scratch to the end. It starts with a cross-validation, performas a model fit, predicts phenotypes and writes complete HTML code with images.


stam.evaluate(expression.matrix, classifications, 
              report.dir = getwd(), aclass = 
              titlestem = NULL, testset = 
              stam.balanced.folds(classifications, 3)[[1]], 
              chip = "hgu95av2", root = "GO:0008150", 
              no.output = FALSE, alpha = seq(0, 1, 0.1), 
              beta = NULL, deltas = NULL, ndeltas = 30, 
              minspec = NULL, minsens = 0.1, maxsens = 1, 
              pamimagefile = NULL)


expression.matrix holds the expression levels. It may be of class exprSet or a plain numeric matrix. In the first case exprs is used to extract the expression levels. The matrix is expected to hold one column per sample and one row per probeset.
classifications This character vector must contain one entry per sample identifying the group it belongs to.
aclass the name of the phenotype class of interest, usually the disease class. If set to NULL the first class in lexicographical order is chosen.
testset indeces of the columns in the expression.matrix representing test samples.
chip the name of the microarray chip. A meta data package of the same name is expected to be found holding the needed annotation, namely the links between probesets and Gene Ontology nodes.
root the GO node used as root of the classifier graph. Only successors of this node are considered during construction of the graph/model.
alpha root performance vs. mean redundancy weight. If set to NULL the root error rate is used exclusively to determine the best shrinkage level. If a numeric vector is provided, all alternatives are computed and the user is given an interactive choice. Values between 0 and 1 are valid, 0 meaning exclusive weight on redundancy and 1 putting exclusive weight on performance.
beta holds class weights used when judging classifier quality. The default is to set class weights to the corresponding prevalence. Several combinations of class weights may be provided for testing one after the other. To do so a matrix is expected to hold one combination of weights per row and must thus have one column per class.
deltas numeric vector holding graph shrinkage candidates. Default is to determine ndelta candidates between 0 and the lowest shrinkage level which removes all leaf nodes.
ndeltas number of automatically determined graph shrinkage candidates determined if deltas is not defined.
titlestem the first part of the title of the HTML page to be written, is complemented by some of the parameters.
report.dir the directory where the HTML pages are to be written
no.output do not generate any HTML or images
minspec nodes to be shown in molecular symptoms image must be at least this specific
minsens nodes to be shown in molecular symptoms image must be at least this sensitive
maxsens nodes to be shown in molecular symptoms image must be at most this sensitive
pamimagefile When this parameter is specified tries to read this file and extract a stamCV object to avoid recomputing PAM fits. If the file does not yet exist, PAM fits are stored there after computation.


stam.evaluate exexutes all steps needed in a structured analysis of a microarray study and coherently generates HTML output including plots and images. In Firstly, a 10 fold cross validation is performed with the data not identified as test set. Secondly, using an adequate graph shrinkage level, a model fit is computed. Finally, all data is used for prediction to illustrate the performance.

Furthermore, this method generates a set of HTML pages. One page reports on the analysis as a whole, while additional interlinked pages, one for each node in the model fit, contain information on the fit and results of each node. On the main page plots and images illustrate and summarize the analysis. Clickable maps make the exploration of the results convenient. All files are stored together with an R data containing the returned R object in the user specified report directory.


Returns an object of class stamEval containing all results generated during the above described procedure. Use the methods defined on the class corresponding the slot you want to investigat further.


Claudio Lottaz

See Also

stamEval-class,,, stam.predict, stam.writeHTML


# load and normalize some data
## Not run: 
golubNorm <- vsn(golubMerge, describe.preprocessing=FALSE)

# (root is chosen to yield results reasonably fast, 
#  consider GO:0008150 (biological process) to obtain 
#  meaningful results)

# demonstrate the use of several combinations of class weights
betas <- cbind(c(0.5, 0.8, 0.9), c(0.5, 0.2, 0.1))
golubNorm.eval.explore <- stam.evaluate(golubNorm, "ALL.AML", 
                                         chip="hu6800", root="GO:0005576", 
                                         alpha=seq(0, 1, 0.1), beta=betas, ndelta=10)

# demonstrate the use of testsets
golubNorm.eval.predict <- stam.evaluate(golubNorm, "ALL.AML", testset=39:72,
                                         chip="hu6800", root="GO:0005576", ndelta=10)
## End(Not run)

[Package stam version 1.0.3 Index]