xMLcaret: Function to integrate predictor matrix in a supervised manner...

In Pi: Leveraging Genetic Evidence to Prioritise Drug Targets at the Gene and Pathway Level

Description

xMLcaret is supposed to integrate predictor matrix in a supervised manner via machine learning algorithms using caret. The caret package streamlines model building and performance evaluation. It requires three inputs: 1) Gold Standard Positive (GSP) targets; 2) Gold Standard Negative (GSN) targets; 3) a predictor matrix containing genes in rows and predictors in columns, with their predictive scores inside it. It returns an object of class 'sTarget'.

Usage

xMLcaret(
list_pNode = NULL,
df_predictor = NULL,
GSP,
GSN,
method = c("gbm", "svmRadial", "rda", "knn", "pls", "nnet", "rf",
"myrf", "cforest",
"glmnet", "glm", "bayesglm", "LogitBoost", "xgbLinear", "xgbTree"),
nfold = 3,
nrepeat = 10,
seed = 825,
aggregateBy = c("none", "logistic", "Ztransform", "fishers",
"orderStatistic"),
verbose = TRUE,
RData.location = "http://galahad.well.ox.ac.uk/bigdata",
guid = NULL
)

Arguments

list_pNode	a list of "pNode" objects or a "pNode" object
df_predictor	a data frame containing genes (in rows) and predictors (in columns), with their predictive scores inside it. This data frame must has gene symbols as row names
GSP	a vector containing Gold Standard Positive (GSP)
GSN	a vector containing Gold Standard Negative (GSN)
method	machine learning method. It can be one of "gbm" for Gradient Boosting Machine (GBM), "svmRadial" for Support Vector Machines with Radial Basis Function Kernel (SVM), "rda" for Regularized Discriminant Analysis (RDA), "knn" for k-nearest neighbor (KNN), "pls" for Partial Least Squares (PLS), "nnet" for Neural Network (NNET), "rf" for Random Forest (RF), "myrf" for customised Random Forest (RF), "cforest" for Conditional Inference Random Forest, "glmnet" for glmnet, "glm" for Generalized Linear Model (GLM), "bayesglm" for Bayesian Generalized Linear Model (BGLM), "LogitBoost" for Boosted Logistic Regression (BLR), "xgbLinear" for eXtreme Gradient Boosting as linear booster (XGBL), "xgbTree" for eXtreme Gradient Boosting as tree booster (XGBT)
nfold	an integer specifying the number of folds for cross validataion. Per fold creates balanced splits of the data preserving the overall distribution for each class (GSP and GSN), therefore generating balanced cross-vallidation train sets and testing sets. By default, it is 3 meaning 3-fold cross validation
nrepeat	an integer specifying the number of repeats for cross validataion. By default, it is 10 indicating the cross-validation repeated 10 times
seed	an integer specifying the seed
aggregateBy	the aggregate method used to aggregate results from repeated cross validataion. It can be either "none" for no aggregration (meaning the best model based on all data used for cross validation is used), or "orderStatistic" for the method based on the order statistics of p-values, or "fishers" for Fisher's method, "Ztransform" for Z-transform method, "logistic" for the logistic method. Without loss of generality, the Z-transform method does well in problems where evidence against the combined null is spread widely (equal footings) or when the total evidence is weak; Fisher's method does best in problems where the evidence is concentrated in a relatively small fraction of the individual tests or when the evidence is at least moderately strong; the logistic method provides a compromise between these two. Notably, the aggregate methods 'Ztransform' and 'logistic' are preferred here
verbose	logical to indicate whether the messages will be displayed in the screen. By default, it sets to TRUE for display
RData.location	the characters to tell the location of built-in RData files. See xRDataLoader for details
guid	a valid (5-character) Global Unique IDentifier for an OSF project. See xRDataLoader for details

Value

an object of class "sTarget", a list with following components:

• model: an object of class "train" as a best model

• ls_model: a list of best models from repeated cross-validation

• priority: a data frame of n X 5 containing gene priority information, where n is the number of genes in the input data frame, and the 5 columns are "GS" (either 'GSP', or 'GSN', or 'Putative'), "name" (gene names), "rank" (priority rank), "rating" (5-star priority score/rating), and "description" (gene description)

• predictor: a data frame, which is the same as the input data frame but inserting two additional columns ('GS' in the first column, 'name' in the second column)

• performance: a data frame of 1+nPredictor X 2 containing the supervised/predictor performance info, where nPredictor is the number of predictors, two columns are "ROC" (AUC values) and "Fmax" (F-max values)

• performance_cv: a data frame of nfold*nrepeat X 2 containing the repeated cross-validation performance, where two columns are "ROC" (AUC values) and "Fmax" (F-max values)

• importance: a data frame of nPredictor X 1 containing the predictor importance info

• gp: a ggplot object for the ROC curve

• gp_cv: a ggplot object for the ROC curves from repeated cross-validation

• evidence: an object of the class "eTarget", a list with following components "evidence" and "metag"

• list_pNode: a list of "pNode" objects

Note

It will depend on whether a package "caret" and its suggested packages have been installed. It can be installed via: BiocManager::install(c("caret","e1071","gbm","kernlab","klaR","pls","nnet","randomForest","party","glmnet","arm","caTools","xgboost")).

See Also

xPierMatrix, xPredictROCR, xPredictCompare, xSparseMatrix, xSymbol2GeneID

Examples

RData.location <- "http://galahad.well.ox.ac.uk/bigdata"
## Not run: 
sTarget <- xMLcaret(df_prediction, GSP, GSN, method="myrf")

## End(Not run)

Pi documentation built on Nov. 26, 2020, 2:01 a.m.