GenoNet.predict: Predict functional consequences of genetic variation based on...

In Ionita-Laza-lab/GenoNet: A Semi-Supervised Approach for Predicting Cell-Type Specific Functional Consequences of Non-Coding Variation

Description

Once the preliminary work is done using "GenoNet()", this function predicts labels for a set of variants given their epigenetic features.

Usage

GenoNet.predict(X,GenoNet.fit)

Arguments

X	Features used for prediction. If pre-calculated parameters from FunDLA are being used, X must have 1016 columns corresponding to the 1016 epigenetic features in 127 tissues or cell types, with the same order. See the complete list of features (ordered) at http://www.funlda.com/genonet/help.
GenoNet.fit	Previously trained GenoNet model.

Value

Y.predict

Predicted outcomes or equivalently labels (quantitative, between 0 and 1).

Examples

## GenoNet trains the model that will be used for prediction.
# Input: Y, X (covariates/features), pre-calculated parameters from unlabeled data

## GenoNet.predict predicts labels given a set of new features.
# Input: X (test data covariates/features), GenoNet.fit (fitted GenoNet model)

library(GenoNet)

#Load example data
data(GenoNet.example)
beta.FunLDA<-GenoNet.example$beta.FunLDA # pre-calculated parameters from unlabeled data
training.data<-GenoNet.example$training.data # example training data
test.data<-GenoNet.example$test.data # example test data

beta.unlabeled<-beta.FunLDA[,'E118'] # tissue E118, HepG2 Hepatocellular Carcinoma Cells
Y.training<-training.data[,4];Y.test<-test.data[,4]
X.training<-training.data[,-c(1:4)];X.test<-test.data[,-c(1:4)]

#fit GenoNet
GenoNet.fit<-GenoNet(Y.training,X.training,beta.unlabeled,nfolds=10)
Y.predict<-GenoNet.predict(X.test,GenoNet.fit)

Ionita-Laza-lab/GenoNet documentation built on Nov. 5, 2019, 2:22 p.m.