MAPK.classifier: A classifier for genetic interaction data.

In Huber-group-EMBL/RNAinteractMAPK: Mapping of Signalling Networks through Synthetic Genetic Interaction Analysis by RNAi

Description

These functions implement a classifier to classify three classes of pathway membership of the RasMAPK and JNK pathway. For each sample and each channel a sparse linear discriminat classifier is trained. The posterior probabilities are averaged over all single classifiers. The classification posterior probabilities of three classes are plotted as a ternary plot (ternary plot adapted from CRAN package vcd).

Usage

MAPK.cv.classifier(sgi, traingroups)
MAPK.predict.classification(sgi, traingroups)
MAPK.plot.classification(posterior,
                         classes = NULL, classnames = NULL, 
		         col = "darkgray", y = NULL,
                         classcol = NULL,
		         main = "predicted classification probabilities",
                         pop = TRUE, threshText = 0.3,
			 textToLeft = NULL, textToRight = NULL)

Arguments

sgi	An object of class RNAinteract
traingroups	A list of gene names for the training examples. For each class there should be a vector of gene names.
posterior	A matrix of posterior probabilities. Each row represents one gene, each column represents one class.
classes	The three classes to be displayed on the ternary plot.
classnames	The class names to be displayed.
col	The color used for the text labels.
y	A factor representing the class label for each gene in posterior.
classcol	The color used for the three classes.
main	The title of the plot.
pop	If TRUE, all viewports are popped before finishing the function.
threshText	A threshold for the posterior probability of the three classes. Only genes that are assigned with a larger probability to the three classes are shown.
textToLeft	These text labels will be shown on the left hand side.
textToRight	These text labels will be shown on the right hand side.

Details

The code for the ternary plot (used by MAPK.plot.classification) is adapted from the function ternaryplot in the CRAN package vcd Author of the original code is David Meyer ([e-mail]). References: M. Friendly (2000), Visualizing Categorical Data. SAS Institute, Cary, NC. This code is specialized for the publication "Mapping Signalling Networks by RNAi ..." in Nat. Methods. It is highly recommended to use the original code by David Meyer.

Value

MAPK.cv.classifier returns a list with the cross validated class assignment probability, as well as the results of the single classifiers.

MAPK.predict.classifier returns the predicted posterior probabilities of new genes as well as the classification results of the single classifier.

MAPK.plot.classifier returns nothing.

Author(s)

Bernd Fischer

References

function ternaryplot, CRAN package vdc. M. Friendly (2000), Visualizing Categorical Data. SAS Institute, Cary, NC.

See Also

RNAinteract, RNAinteractMAPK

Huber-group-EMBL/RNAinteractMAPK documentation built on Dec. 31, 2020, 1:01 p.m.