svmOnto: Ontology-term predictions by binary SVM classifiers

In fspetale/fgga: Hierarchical ensemble method based on factor graph

Ontology-term predictions by binary SVM classifiers

Description

svmOnto delivers soft Ontology-term predictions based on binary SVM classification models.

Usage

svmOnto(svmMoldel, dxCharacterized, rootNode, varianceSVM)

Arguments

svmMoldel	A list of object of class “svm" created by svm.
dxCharacterized	A data frame with ‘n’ protein coding genes (rows) by ‘f’ features (columns).
rootNode	A character indicating the root of the graph.
varianceSVM	A vector named with the variance of cross-Ontology node labels.

Details

Binary SVM predictions are supplemented with their corresponding margins. These margins are used to model the additive zero-mean Gaussian learning noise that corrupts ideal but hidden Ontology-term predictions. These ideal predictions are embedded in hidden variable nodes of the Forney Factor Graph.

Value

svmOnto

A named vector of predicted values for a protein sequence.

Author(s)

Flavio E. Spetale, Pilar Bulacio and Javier Murillo <[e-mail]>

References

Chang, Chih-Chung and Lin, Chih-Jen: LIBSVM: a library for Support Vector Machines http://www.csie.ntu.edu.tw/~cjlin/libsvm

Eisner R, Poulin B, Szafron D, Lu P, Greiner R. Improving protein function prediction using the hierarchical structure of the Gene Ontology. In: Proc. IEEE CIBCB; 2005. p. 1–1

Spetale FE, Tapia E, Krsticevic F, Roda F, Bulacio P (2016). A Factor Graph Approach to Automated GO Annotation. PLOS ONE 11(1): e0146986

See Also

svmTrain

Examples

data(CfData)

mygraphGO <- as(CfData[["graphCfGO"]], "graphNEL")

modelSVMs <- lapply(CfData[["nodesGO"]][1:4], FUN = svmTrain,
                    tableOntoTerms = CfData[["tableCfGO"]],
                    dxCharacterized = CfData[["dxCf"]],
                    graphOnto = mygraphGO, kernelSVM = "radial")

rootGO <- leaves(mygraphGO, "in")

varianceGOs <- CfData[["varianceGOs"]]

# SVM testing in four GO-terms
dxTestCharacterized <- CfData[["dxCf"]][
                                sample(1:dim(CfData[["dxCf"]])[1], 20), ]

matrixGOTest <- svmOnto(svmMoldel = modelSVMs,
                        dxCharacterized = dxTestCharacterized,
                        rootNode = rootGO, varianceSVM = varianceGOs)

fspetale/fgga documentation built on Jan. 29, 2024, 6:53 p.m.