predictorKMR: Wrapper function for kernel multitask regression

In jpvert/kmr4toxicogenetics: Kernel multitask regression for toxicogenetics

Description

Wrapper function to perform kernel multitask regression with cv.kmr that trains a model on training set and then predicts on test set for multiple tasks.

Usage

predictorKMR(patientsKernelTrain, patientsKernelTest, response, drugsKernel,
  lambdas = exp(-15:25), nfolds = 5, nrepeats = 1)

Arguments

patientsKernelTrain	Precomputed kernel Gram matrix of n training patients, of dimension n x n.
patientsKernelTest	Precomputed kernel Gram matrix of m test patients crossing n training patients, of dimension m x n.
response	Matrix of observed toxicity values, of dimension n x t, for the n training patients responding to t drugs.
drugsKernel	Kernel Gram matrix of the t drugs, of dimension t x t.
lambdas	Sequence of lambdas that must be tested to fit a cross-validated KMR model. Default is exp(-15:25).
nfolds	Number of folds for cross-validation. Default is 5.
nrepeats	Number of times the k-fold cross-validation is performed. Default is 1.

Value

A matrix of predicted toxicity values, of dimension m x t, for the m test patients responding to the t drugs.

Note

Multitask prediction is made, for which task relationships are encoded in drugsKernel.

References

Bernard, E., Jiao, Y., Scornet, E., Stoven, V., Walter, T., and Vert, J.-P. (2017). "Kernel multitask regression for toxicogenetics." bioRxiv-171298.

See Also

cv.kmr

jpvert/kmr4toxicogenetics documentation built on May 24, 2019, 2:04 a.m.