Type: Package
Title: Fusing Machine Learning in R
Version: 0.0.1
Maintainer: Cesaire J. K. Fouodo <cesaire.kuetefouodo@uni-luebeck.de>
Description: Recent technological advances have enable the simultaneous collection of multi-omics data i.e., different types or modalities of molecular data, presenting challenges for integrative prediction modeling due to the heterogeneous, high-dimensional nature and possible missing modalities of some individuals. We introduce this package for late integrative prediction modeling, enabling modality-specific variable selection and prediction modeling, followed by the aggregation of the modality-specific predictions to train a final meta-model. This package facilitates conducting late integration predictive modeling in a systematic, structured, and reproducible way.
License: GPL-3
Encoding: UTF-8
Imports: R6, stats, digest
Suggests: testthat (≥ 3.0.0), UpSetR (≥ 1.4.0), caret, ranger, glmnet, Boruta, knitr, rmarkdown, pROC, checkmate
Config/testthat/edition: 3
Depends: R (≥ 3.6.0)
RoxygenNote: 7.3.2
Collate: 'Data.R' 'HashTable.R' 'Lrner.R' 'Model.R' 'PredictData.R' 'PredictLayer.R' 'PredictMetaLayer.R' 'Predicting.R' 'Target.R' 'TestData.R' 'TestLayer.R' 'TestMetaLayer.R' 'Testing.R' 'TrainData.R' 'TrainLayer.R' 'TrainMetaLayer.R' 'Training.R' 'VarSel.R' 'bestLayerLearner.R' 'cobra.R' 'createCobraPred.R' 'createDif.R' 'createLoss.R' 'createWeights.R' 'multi_omics.R' 'predict.bestLayerLearner.R' 'predict.cobra.R' 'weightedMeanLearner.R' 'predict.weightedMeanLearner.R' 'testingFunctions.R' 'trainingFunctions.R'
VignetteBuilder: knitr, rmarkdown
BugReports: https://github.com/imbs-hl/fuseMLR/issues
NeedsCompilation: no
Packaged: 2024-12-15 04:36:16 UTC; fouodo
Author: Cesaire J. K. Fouodo [aut, cre]
Repository: CRAN
Date/Publication: 2024-12-17 09:20:02 UTC

Abstract class Data

Description

As abstract, a Data object cannot be stored on any layer. Instead, extended TrainData or TestData objects can be stored on a layer.

Methods

Public methods

Method new()

Constructor of class Data.

Usage
Data$new(id, ind_col, data_frame)
Arguments
id

character
Object ID.

ind_col

character
Column name containing individual IDs.

data_frame

data.frame
data.frame containing data.

Method print()

Printer

Usage
Data$print(...)
Arguments
...

any

Method getIndSubset()

Retrieve a data subset for a given variable name and values, a data subset.

Usage
Data$getIndSubset(var_name, value)
Arguments
var_name

character
Variable name of interest.

value

vector
Values of interest.

Returns

The data subset is returned.

Method impute()

Imputes missing values in modality-specific predictions. Only mode and median based imputations are actually supported.

Usage
Data$impute(impute_fct, impute_param, target_name)
Arguments
impute_fct

character
An imputation function to use instead of median or mode imputation. Not yet implemented!

impute_param

list

target_name

character
Name of the target variable. The list of parameters to call the imputation function.

Returns

A new object with the predicted values is returned.

Method getVarSubset()

Retrieve a subset of variables from data.

Usage
Data$getVarSubset(var_name)
Arguments
var_name

character
Variable names of interest.

Returns

The data subset is returned.

Method getSetDiff()

For the given variable name, non existing values in the current dataset are returned.

Usage
Data$getSetDiff(var_name, value)
Arguments
var_name

character
Variable name of interest.

value

vector
Values of interest.

Returns

The subset difference is returned.

Method getDataFrame()

Getter of the data.frame.

Usage
Data$getDataFrame()
Returns

The data.frame of the current object is returned.

Method setDataFrame()

Set a new data.frame to the current object.

Usage
Data$setDataFrame(data_frame)
Arguments
data_frame

data.frame

Returns

The current object is returned.

Method getCompleteData()

Getter of the complete dataset without missing values.

Usage
Data$getCompleteData()
Returns

The complete dataset is returned.

Method getId()

Getter of the current object ID.

Usage
Data$getId()
Returns

The current object ID is returned.

Method getData()

Getter of the current Data. This function is re-implemented by TrainData and TestData.

Usage
Data$getData()
Returns

Do not use on this class.

Method getIndCol()

Getter of the individual column variable.

Usage
Data$getIndCol()

Method clone()

The objects of this class are cloneable with this method.

Usage
Data$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

See Also

TrainData and TestData

Class HashTable

Description

Hashtable to contain object modalities. Storage objects like Training and TrainLayer are extensions of this class.

Methods

Public methods

Method new()

Initialize a default parameters list.

Usage
HashTable$new(id)
Arguments
id

character
ID of the hash table. It must be unique.

Method print()

Printer

Usage
HashTable$print(...)
Arguments
...

any

Method add2HashTable()

Function to add a key-value pair to the hash table.

Usage
HashTable$add2HashTable(key, value, .class)
Arguments
key

character
The key to be added.

value

object
Object to be added.

.class

character
Class of the object to be added.

Method getFromHashTable()

Getter of the object which the key passed as argument.

Usage
HashTable$getFromHashTable(key)
Arguments
key

character
Key of the required object.

Method getKeyClass()

Getter of the data.frame that stores all key class pairs.

Usage
HashTable$getKeyClass()
Returns

data.frame

Method removeFromHashTable()

Remove the object with the corresponding key from the hashtable.

Usage
HashTable$removeFromHashTable(key)
Arguments
key

Key of the object to be removed.

Method getId()

Getter of the current object ID.

Usage
HashTable$getId()

Method getHashTable()

Getter of the current hashtable.

Usage
HashTable$getHashTable()

Method checkClassExist()

Check whether object from a class has already been stored.

Usage
HashTable$checkClassExist(.class)
Arguments
.class

character

Returns

Boolean value

Lrner Class

Description

This class implements a learner. A Lrner object can only exist as a component of a TrainLayer or a TrainMetaLayer object.

Methods

Public methods

Method new()

Initialize a default parameters list.

Usage
Lrner$new(
  id,
  package = NULL,
  lrn_fct,
  param_train_list,
  param_pred_list = list(),
  train_layer,
  na_action = "na.rm"
)
Arguments
id

character
Learner ID.

package

character
Package that implements the learn function. If NULL, the

lrn_fct

character
learn function is called from the current environment.

param_train_list

list
List of parameter for training.

param_pred_list

list
List of parameter for testing. Learn parameters.

train_layer

TrainLayer
Layer on which the learner is stored.

na_action

character
Handling of missing values. Set to "na.keep" to keep missing values, "na.rm" to remove individuals with missing values or "na.impute" (only applicable on meta-data) to impute missing values in meta-data. Only median and mode based imputations are actually handled. With the "na.keep" option, ensure that the provided learner can handle missing values.

Method print()

Printer

Usage
Lrner$print(...)
Arguments
...

any

Method summary()

Printer

Usage
Lrner$summary(...)
Arguments
...

any

Method interface()

Learner and prediction parameter interface. Use this function to provide how the following parameters are named in the learning function (lrn_fct) you provided when creating the learner, or in the predicting function.

Usage
Lrner$interface(
  x = "x",
  y = "y",
  object = "object",
  data = "data",
  extract_pred_fct = NULL
)
Arguments
x

character
Name of the argument to pass the matrix of independent variables in the original learning function.

y

character
Name of the argument to pass the response variable in the original learning function.

object

character
Name of the argument to pass the model in the original predicting function.

data

character
Name of the argument to pass new data in the original predicting function.

extract_pred_fct

character or function
If the predict function that is called for the model does not return a vector, then use this argument to specify a (or a name of a) function that can be used to extract vector of predictions. Default value is NULL, if predictions are in a vector.

Method train()

Tains the current learner (from class Lrner) on the current training data (from class TrainData).

Usage
Lrner$train(ind_subset = NULL, use_var_sel = FALSE, verbose = TRUE)
Arguments
ind_subset

vector
Individual ID subset on which the training will be performed.

use_var_sel

boolean
If TRUE, variable selection is performed before training.

verbose

boolean
Warning messages will be displayed if set to TRUE.

Returns

The resulting model, from class Model, is returned.

Method getTrainLayer()

The current layer is returned.

Usage
Lrner$getTrainLayer()
Returns

TrainLayer object.

Method getNaRm()

The current layer is returned.

Usage
Lrner$getNaRm()

Method getNaAction()

The current layer is returned.

Usage
Lrner$getNaAction()

Method getId()

Getter of the current learner ID.

Usage
Lrner$getId()
Returns

The current learner ID.

Method getPackage()

Getter of the learner package implementing the learn function.

Usage
Lrner$getPackage()
Returns

The name of the package implementing the learn function.

Method getIndSubset()

Getter of the learner package implementing the learn function.

Usage
Lrner$getIndSubset()
Returns

The name of the package implementing the learn function.

Method getVarSubset()

Getter of the variable subset used for training.

Usage
Lrner$getVarSubset()
Returns

The list of variables used for training is returned.

Method getParamPred()

Getter predicting parameter list.

Usage
Lrner$getParamPred()
Returns

The list of predicting parameters.

Method getParamInterface()

The current parameter interface is returned.

Usage
Lrner$getParamInterface()
Returns

A data.frame of interface.

Method getExtractPred()

The function to extract predicted values is returned.

Usage
Lrner$getExtractPred()
Returns

A data.frame of interface.

Model Class

Description

This class implements a model. A Model object can only exist as element of a TrainLayer or a TrainMetaLayer object. A Model object is automatically created by fitting a learner on a training data.

A Model object can compute predictions for a TestData object. See the predict function below.

Methods

Public methods

Method new()

Constructor of Model class.

Usage
Model$new(lrner, train_data, base_model, train_layer)
Arguments
lrner

Lrner
The learner.

train_data

TrainData(1)
Training data.

base_model

object
Base model as returned by the original learn function.

train_layer

TrainLayer
The current training layer on which the model is stored.

Returns

An object is returned.

Method print()

Printer

Usage
Model$print(...)
Arguments
...

any

Method summary()

Summary

Usage
Model$summary(...)
Arguments
...

any

Method getBaseModel()

Getter of the base model

Usage
Model$getBaseModel()

Method getTrainData()

Getter of the traning data

Usage
Model$getTrainData()

Method getTrainLabel()

Getter of the individual ID column in the training data.

Usage
Model$getTrainLabel()
Arguments
...

any

Method getLrner()

Getter of the learner use to fit the model.

Usage
Model$getLrner()

Method setId()

Setter of the model ID.

Usage
Model$setId(id)
Arguments
id

character
ID value

Method predict()

Predict target values for the new data (from class TestData) taken as into.

Usage
Model$predict(testing_data, use_var_sel, ind_subset = NULL)
Arguments
testing_data

TestData
An object from class TestData.

use_var_sel

boolean
If TRUE, selected variables available at each layer are used.

ind_subset

vector
Subset of individual IDs to be predicted.

...

Further parameters to be passed to the basic predict function.

Returns

The predicted object are returned. The predicted object must be either a vector or a list containing a field predictions with predictions.

Method clone()

The objects of this class are cloneable with this method.

Usage
Model$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

PredictData Class

Description

This class implements PredictData object to be predicted. A PredictData object can only exist as a component of a PredictLayer or a PredictMetaLayer object.

Super class

fuseMLR::Data -> PredictData

Methods

Public methods

Inherited methods

Method new()

Initialize a new object from the current class.

Usage
PredictData$new(id, ind_col, data_frame)
Arguments
id

character
Object ID.

ind_col

character
Column name containing individual IDs.

data_frame

data.frame
data.frame containing data.

Method print()

Printer

Usage
PredictData$print(...)
Arguments
...

any

Method getPredictData()

Getter of the current predicted data.frame wihtout individual ID variable.

Usage
PredictData$getPredictData()
Returns

The data.frame without individual ID nor target variables is returned.

Method getPredictLayer()

Getter of the current layer.

Usage
PredictData$getPredictLayer()
Returns

The layer (from class PredictLayer) on which the current train data are stored is returned.

Method setPredictLayer()

Assigns a predicted layer to the predicted data.

Usage
PredictData$setPredictLayer(predict_layer)
Arguments
predict_layer

PredictLayer(1)

Returns

The current object

Method clone()

The objects of this class are cloneable with this method.

Usage
PredictData$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

See Also

TrainData, TestData

PredictLayer Class

Description

This class implements a layer. A PredictLayer object can only exist as a component of a Predicting object.

A predicted layer can only contain PredictData.

Super class

fuseMLR::HashTable -> PredictLayer

Methods

Public methods

Inherited methods

Method new()

constructor

Usage
PredictLayer$new(id)
Arguments
id

character
The layer ID.

Method print()

Printer

Usage
PredictLayer$print(...)
Arguments
...

any

Method getPredicting()

Getter of the current predicting object

Usage
PredictLayer$getPredicting()
Returns

The current predicting object is returned.

Method getIndIDs()

Getter of IDS from the current layer.

Usage
PredictLayer$getIndIDs()
Returns

A data.frame containing individuals IDs values.

Method getPredictData()

Getter of the predicted data stored on the current layer.

Usage
PredictLayer$getPredictData()
Returns

The stored PredictData object is returned.

Method setPredicting()

Assigns a predicting object to the predicted layer.

Usage
PredictLayer$setPredicting(predicting)
Arguments
predicting

Predicting

Returns

The current object

Method summary()

Generate summary.

Usage
PredictLayer$summary()

See Also

Training, Lrner, TrainData, TestData and Model

PredictMetaLayer Class

Description

This class implement a predicted meta layer. A PredictMetaLayer can only exist as unique element of a Training object.

A predicted meta layer can only contain a PredictData object.

Super class

fuseMLR::HashTable -> PredictMetaLayer

Methods

Public methods

Inherited methods

Method new()

constructor

Usage
PredictMetaLayer$new(id, predicting)
Arguments
id

character

predicting

Predicting

Method print()

Printer

Usage
PredictMetaLayer$print(...)
Arguments
...

any

Method getPredicting()

Getter of the current predicting object

Usage
PredictMetaLayer$getPredicting()
Returns

The current predicting object is returned.

Method getIndIDs()

Getter of IDS from the current layer.

Usage
PredictMetaLayer$getIndIDs()
Returns

A data.frame containing individuals IDs values.

Method getPredictData()

Getter of the predicted data.

Usage
PredictMetaLayer$getPredictData()
Returns

The stored PredictData object is returned.

Method openAccess()

Open access to the meta layer. A meta learner is only modifiable if the access is opened.

Usage
PredictMetaLayer$openAccess()

Method closeAccess()

Close access to the meta layer to avoid accidental modification.

Usage
PredictMetaLayer$closeAccess()

Method getAccess()

Getter of the current access to the meta layer.

Usage
PredictMetaLayer$getAccess()

Predicting Class

Description

This class is designed for predictions.

The Predicting is structured as followed:

Use the function train for training and predict for predicting.

TODO: Do not export me.

Super class

fuseMLR::HashTable -> Predicting

Methods

Public methods

Inherited methods

Method new()

constructor

Usage
Predicting$new(id, ind_col)
Arguments
id

character
Predicting id.

ind_col

character Name of column of individuals IDS

Method print()

Printer

Usage
Predicting$print(...)
Arguments
...

any

Method createMetaTestData()

Creates a new modality-specific predictions dataset based on layer predictions.

Usage
Predicting$createMetaTestData(meta_layer_id)
Arguments
meta_layer_id

(character(1))
ID of the meta layer where the testing meta data will be stored.

Returns

A TestData is returned.

Method getIndIDs()

Gather individual IDs from all layer.

Usage
Predicting$getIndIDs()
Returns

A data.frame containing individuals IDs.

Method getPredictMetaLayer()

Getter of the meta layer.

Usage
Predicting$getPredictMetaLayer()
Returns

Object from class PredictMetaLayer

Method getIndCol()

Getter of the individual column name.

Usage
Predicting$getIndCol()

See Also

TrainLayer

Target Class

Description

This class implements the target object. A Target object can only exist as a component of a Training object.

Super class

fuseMLR::Data -> Target

Methods

Public methods

Inherited methods

Method new()

Initialize a new object from the current class.

Usage
Target$new(id, data_frame, training)
Arguments
id

character
The Object ID.

data_frame

data.frame
data.frame containing data.

training

Training
Training where to store the current object.

Method print()

Printer

Usage
Target$print(...)
Arguments
...

any

Method summary()

Summary

Usage
Target$summary(...)
Arguments
...

any

Method getData()

Getter of the current data.frame wihtout individual ID nor target variables.

Usage
Target$getData()
Returns

The data.frame without individual ID nor target variables is returned.

Method getTargetValues()

Getter of target values stored on the current training layer.

Usage
Target$getTargetValues()
Returns

The observed target values stored on the current training layer are returned.

Method getTargetName()

Getter of the target variable name.

Usage
Target$getTargetName()

Method getTraining()

Getter of the current training object.

Usage
Target$getTraining()
Returns

The training layer (from class Training) on which the current train data are stored is returned.

Method setData()

Getter of the current data.frame wihtout individual ID nor target variables.

Usage
Target$setData(data_frame)
Arguments
data_frame

data.frame
data.frame to be set.

Title

Method clone()

The objects of this class are cloneable with this method.

Usage
Target$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

See Also

TrainLayer, Lrner, Model, TestData

TestData Class

Description

This class implements TestData object to be predicted. A TestData object can only exist as a component of a TestLayer or a TestMetaLayer object.

Super class

fuseMLR::Data -> TestData

Methods

Public methods

Inherited methods

Method new()

Initialize a new object from the current class.

Usage
TestData$new(id, data_frame, new_layer)
Arguments
id

character
Object ID.

data_frame

data.frame
data.frame containing data.

new_layer

TestLayer
Layer where to store the current object.

ind_col

character
Column name containing individual IDs.

Method print()

Printer

Usage
TestData$print(...)
Arguments
...

any

Method getData()

Getter of the current data.frame wihtout individual ID variable.

Usage
TestData$getData()
Returns

The data.frame without individual ID nor target variables is returned.

Method getTestLayer()

Getter of the current layer.

Usage
TestData$getTestLayer()
Returns

The layer (from class TestLayer) on which the current train data are stored is returned.

Method clone()

The objects of this class are cloneable with this method.

Usage
TestData$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

See Also

TrainData

TestLayer Class

Description

This class implements a layer. A TestLayer object can only exist as a component of a Predicting object.

A predicted layer can only contain TestData.

Super class

fuseMLR::HashTable -> TestLayer

Methods

Public methods

Inherited methods

Method new()

constructor

Usage
TestLayer$new(id, testing)
Arguments
id

character
Testing layer id.

testing

Testing

Method print()

Printer

Usage
TestLayer$print(...)
Arguments
...

any

Method getTesting()

Getter of the current Testing object.

Usage
TestLayer$getTesting()
Returns

The current Testing object is returned.

Method getIndIDs()

Getter of IDS from the current layer.

Usage
TestLayer$getIndIDs()
Returns

A data.frame containing individuals IDs values.

Method getTestData()

Getter of the predicted data stored on the current layer.

Usage
TestLayer$getTestData()
Returns

The stored TestData object is returned.

Method checkTestDataExist()

Check whether a new data has been already stored.

Usage
TestLayer$checkTestDataExist()
Returns

Boolean value

Method summary()

Generate summary.

Usage
TestLayer$summary()

See Also

Training, Lrner, TrainData, TestData and Model

TestMetaLayer Class

Description

This class implement a predicted meta layer. A TestMetaLayer can only exist as unique element of a Training object.

A predicted meta layer can only contain a TestData object.

Super class

fuseMLR::HashTable -> TestMetaLayer

Methods

Public methods

Inherited methods

Method new()

constructor

Usage
TestMetaLayer$new(id, testing)
Arguments
id

character
Testing meta-layer id.

testing

Testing

Method print()

Printer

Usage
TestMetaLayer$print(...)
Arguments
...

any

Method getTesting()

Getter of the current testing object.

Usage
TestMetaLayer$getTesting()
Returns

The current testing object is returned.

Method getTestData()

Getter of the training dataset stored on the current layer.

Usage
TestMetaLayer$getTestData()
Returns

The stored TestData object is returned.

Method openAccess()

Open access to the meta layer. A meta learner is only modifiable if the access is opened.

Usage
TestMetaLayer$openAccess()

Method closeAccess()

Close access to the meta layer to avoid accidental modification.

Usage
TestMetaLayer$closeAccess()

Method getAccess()

Getter of the current access to the meta layer.

Usage
TestMetaLayer$getAccess()

Method setTestData()

Create and set an TestData object to the current new meta learner.

Usage
TestMetaLayer$setTestData(id, ind_col, data_frame)
Arguments
id

character(1)
ID of the TestData object to be instanciated.

ind_col

character(1)
Name of individual column IDs.

data_frame

data.frame(1)
data.frame of layer specific predictions.

Method checkTestDataExist()

Check whether a new data has been already stored.

Usage
TestMetaLayer$checkTestDataExist()
Returns

Boolean value

Testing Class

Description

This is a primary classes of fuseMLR. An object from this class is designed to contain multiple layers, but only one new meta layer.

A Testing object is structured as followed:

Super class

fuseMLR::HashTable -> Testing

Methods

Public methods

Inherited methods

Method new()

constructor

Usage
Testing$new(id, ind_col, verbose = TRUE)
Arguments
id

character
Testing id.

ind_col

character Name of column of individuals IDS in testing data.frame.

verbose

boolean
Warning messages will be displayed if set to TRUE.

Method print()

Printer

Usage
Testing$print(...)
Arguments
...

any

Method getIndIDs()

Gather individual IDs from all layer.

Usage
Testing$getIndIDs()
Returns

A data.frame containing individuals IDs.

Method getTestMetaLayer()

Getter of the meta layer.

Usage
Testing$getTestMetaLayer()
Returns

Object from class TestMetaLayer

Method getIndCol()

Getter of the individual column name.

Usage
Testing$getIndCol()

Method getVerbose()

Getter of the verbose setting.

Usage
Testing$getVerbose()

Method getData()

Retrieve modality-specific prediction data.

Usage
Testing$getData()
Returns

A list containing all (base and meta) models.

Method upset()

UpSet plot to show an overview of the overlap of individuals across various layers.

Usage
Testing$upset(...)
Arguments
...

any
Further parameters to be passed to the the upset function from package UpSetR.

Method summary()

Generate testing summary

Usage
Testing$summary()

See Also

TrainLayer

TrainData Class

Description

This class implements the training data. A TrainData object can only exist as a component of a TrainLayer or a TrainMetaLayer object.

Super class

fuseMLR::Data -> TrainData

Methods

Public methods

Inherited methods

Method new()

Initialize a new object from the current class.

Usage
TrainData$new(id, data_frame, train_layer)
Arguments
id

character
The Object ID.

data_frame

data.frame
data.frame containing data.

train_layer

TrainLayer
Training layer where to store the current object.

Method print()

Printer

Usage
TrainData$print(...)
Arguments
...

any

Method summary()

Summary

Usage
TrainData$summary(...)
Arguments
...

any

Method getData()

Getter of the current data.frame wihtout individual ID nor target variables.

Usage
TrainData$getData()
Returns

The data.frame without individual ID nor target variables is returned.

Method getTargetValues()

Getter of target values stored on the current training layer.

Usage
TrainData$getTargetValues()
Returns

The observed target values stored on the current training layer are returned.

Method getTargetName()

Getter of the target variable name.

Usage
TrainData$getTargetName()

Method getTrainLayer()

Getter of the current training layer.

Usage
TrainData$getTrainLayer()
Returns

The training layer (from class TrainLayer) on which the current train data are stored is returned.

Method getTestLayer()

Getter of the current layer.

Usage
TrainData$getTestLayer()
Returns

The layer (from class TestLayer) on which the current train data are stored is returned.

Method setDataFrame()

Set a new data.frame to the current object.

Usage
TrainData$setDataFrame(data_frame)
Arguments
data_frame

data.frame

Returns

The current object is returned.

Method clone()

The objects of this class are cloneable with this method.

Usage
TrainData$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

See Also

TrainLayer, Lrner, Model, TestData

TrainLayer Class

Description

This class implements a traning layer. A TrainLayer object can only exist as a component of a Training object.

A training layer is structured as followed:

A training layer can train its learner on its training data and store the resulting model. See the public function Layer$train() below.

A training layer can make predictions for a new layer passed as argument to its predict function. See the public function Layer$predict() below.

Super class

fuseMLR::HashTable -> TrainLayer

Methods

Public methods

Inherited methods

Method new()

constructor

Usage
TrainLayer$new(id, training)
Arguments
id

character
Training layer id.

training

Training

Method print()

Printer

Usage
TrainLayer$print(...)
Arguments
...

any

Method getTraining()

Getter of the current training object.

Usage
TrainLayer$getTraining()
Returns

The current training object is returned.

Method getTargetObj()

Getter of the target object.

Usage
TrainLayer$getTargetObj()

Method train()

Trains the current layer.

Usage
TrainLayer$train(ind_subset = NULL, use_var_sel = FALSE, verbose = TRUE)
Arguments
ind_subset

vector
ID subset of individuals to be used for training.

use_var_sel

boolean
If TRUE, variable selection is performed before training.

verbose

boolean
Warning messages will be displayed if set to TRUE.

Returns

The current layer is returned with the resulting model.

Method varSelection()

Variable selection on the current layer.

Usage
TrainLayer$varSelection(ind_subset = NULL, verbose = TRUE)
Arguments
ind_subset

vector
ID subset of individuals to be used for variable selection.

verbose

boolean
Warning messages will be displayed if set to TRUE.

Returns

The current layer is returned with the resulting model.

Method predict()

Predicts values for the new layer taking as argument.

Usage
TrainLayer$predict(new_layer, use_var_sel, ind_subset = NULL)
Arguments
new_layer

TrainLayer

use_var_sel

boolean
If TRUE, selected variables available at each layer are used.

ind_subset

vector

Returns

A new PredictLayer object with the predicted data is returned.

Method getTrainData()

Getter of the training dataset stored on the current layer.

Usage
TrainLayer$getTrainData()
Returns

The stored TrainData object is returned.

Method getTargetValues()

Getter of target values from the current layer.

Usage
TrainLayer$getTargetValues()
Returns

A data.frame containing individuals IDs and corresponding target values.

Method getIndIDs()

Getter of IDS from the current layer.

Usage
TrainLayer$getIndIDs()
Returns

A data.frame containing individuals IDs values.

Method getTestData()

Getter of the new data.

Usage
TrainLayer$getTestData()
Returns

The stored TestData object is returned.

Method getLrner()

Getter of the learner.

Usage
TrainLayer$getLrner()
Returns

The stored Lrner object is returned.

Method getVarSel()

Getter of the variable selector.

Usage
TrainLayer$getVarSel()
Returns

The stored VarSel object is returned.

Method getModel()

Getter of the model.

Usage
TrainLayer$getModel()
Returns

The stored Model object is returned.

Method checkLrnerExist()

Check whether a learner has been already stored.

Usage
TrainLayer$checkLrnerExist()
Returns

Boolean value

Method checkModelExist()

Check whether a model has been already stored.

Usage
TrainLayer$checkModelExist()
Returns

Boolean value

Method checkVarSelExist()

Check whether a variable selection tool has been already stored.

Usage
TrainLayer$checkVarSelExist()
Returns

Boolean value

Method checkTrainDataExist()

Check whether a training data has been already stored.

Usage
TrainLayer$checkTrainDataExist()
Returns

Boolean value

Method summary()

Generate summary.

Usage
TrainLayer$summary()

See Also

Training, Lrner, TrainData, TestData and Model

TrainMetaLayer Class

Description

This class implement a meta meta layer. A TrainMetaLayer can only exist as unique element of a Training object.

A layer is structured as followed:

A meta layer can train its meta learner on the meta training data and store the resulting meta model. The meta layer can predict values given a new meta layer.

Super class

fuseMLR::HashTable -> TrainMetaLayer

Methods

Public methods

Inherited methods

Method new()

constructor

Usage
TrainMetaLayer$new(id, training)
Arguments
id

character
Id of training meta-layer.

training

Training

Method print()

Printer

Usage
TrainMetaLayer$print(...)
Arguments
...

any

Method getTraining()

Getter of the current training object.

Usage
TrainMetaLayer$getTraining()
Returns

The current training object is returned.

Method getTargetObj()

Getter of the target object.

Usage
TrainMetaLayer$getTargetObj()

Method train()

Trains the current layer.

Usage
TrainMetaLayer$train(ind_subset = NULL, verbose = TRUE)
Arguments
ind_subset

vector
ID subset of individuals to be used for training.

verbose

boolean
Warning messages will be displayed if set to TRUE.

Returns

The current layer is returned with the resulting model.

Method predict()

Predicts values for the new layer taking as argument.

Usage
TrainMetaLayer$predict(new_layer, ind_subset = NULL)
Arguments
new_layer

TrainLayer
A trained TrainLayer object.

ind_subset

vector
Index subset.

Returns

A new object with the predicted values is returned.

Method impute()

Imputes missing values in modality-specific predictions. Only mode and median based imputations are actually supported.

Usage
TrainMetaLayer$impute(impute_fct = NULL, impute_param = NULL)
Arguments
impute_fct

character
An imputation function to use instead of median or mode imputation. This parameter is actually not used. This corresponds to median or mode based imputation.

impute_param

list
The list of parameters to call the imputation function. Not yet implemented!

Returns

A new object with the predicted values is returned.

Method getTrainData()

Getter of the training dataset stored on the current layer.

Usage
TrainMetaLayer$getTrainData()
Returns

The stored TrainData object is returned.

Method getLrner()

Getter of the learner.

Usage
TrainMetaLayer$getLrner()
Returns

The stored Lrner object is returned.

Method getModel()

Getter of the model.

Usage
TrainMetaLayer$getModel()
Returns

The stored Model object is returned.

Method openAccess()

Open access to the meta layer. A meta learner is only modifiable if the access is opened.

Usage
TrainMetaLayer$openAccess()

Method closeAccess()

Close access to the meta layer to avoid accidental modification.

Usage
TrainMetaLayer$closeAccess()

Method getAccess()

Getter of the current access to the meta layer.

Usage
TrainMetaLayer$getAccess()

Method setTrainData()

Create and set an TrainData object to the current meta learner.

Usage
TrainMetaLayer$setTrainData(id, ind_col, data_frame)
Arguments
id

character
ID of the TrainData object to be instanciated.

ind_col

character
Name of individual column IDs.

data_frame

data.frame
data.frame of layer specific predictions.

Method checkLrnerExist()

Check whether a training data has been already stored.

Usage
TrainMetaLayer$checkLrnerExist()
Returns

Boolean value

Method checkModelExist()

Check whether a model has been already stored.

Usage
TrainMetaLayer$checkModelExist()
Returns

Boolean value

Method checkTrainDataExist()

Check whether a training data has been already stored.

Usage
TrainMetaLayer$checkTrainDataExist()
Returns

Boolean value

Method set2NotTrained()

Only usefull to reset status FALSE after cross validation.

Usage
TrainMetaLayer$set2NotTrained()

Method summary()

Generate summary.

Usage
TrainMetaLayer$summary()

Training Class

Description

This is a primary classes of fuseMLR. An object from this class is designed to contain multiple training layers, but only one meta training layer.

The Training class is structured as followed:

Use the function train for training and predict for predicting.

Super class

fuseMLR::HashTable -> Training

Methods

Public methods

Inherited methods

Method new()

constructor

Usage
Training$new(
  id,
  ind_col,
  target,
  target_df,
  problem_type = "classification",
  verbose = TRUE
)
Arguments
id

character

ind_col

character
Name of column of individuals IDS.

target

character
Name of the target variable.

target_df

data.frame
Data frame with two columns: individual IDs and response variable values.

problem_type

character
Either "classification" or "regression".

verbose

boolean
Warning messages will be displayed if set to TRUE.

Method print()

Printer

Usage
Training$print(...)
Arguments
...

any

Method trainLayer()

Train each layer of the current Training.

Usage
Training$trainLayer(ind_subset = NULL, use_var_sel = FALSE, verbose = TRUE)
Arguments
ind_subset

character
Subset of individuals IDs to be used for training.

use_var_sel

boolean
If TRUE, selected variables available at each layer are used.

verbose

boolean
Warning messages will be displayed if set to TRUE.

Returns

Returns the object itself, with a model for each layer.

Method predictLayer()

Predicts values given new data.

Usage
Training$predictLayer(testing, ind_subset = NULL)
Arguments
testing

TestData
Object of class TestData.

ind_subset

vector
Subset of individuals IDs to be used for training.

Returns

A new Training with predicted values for each layer.

Method createMetaTrainData()

Creates a meta training dataset and assigns it to the meta layer.

Usage
Training$createMetaTrainData(
  resampling_method,
  resampling_arg,
  use_var_sel,
  impute = TRUE
)
Arguments
resampling_method

function
Function for internal validation.

resampling_arg

list
List of arguments to be passed to the function.

use_var_sel

boolean
If TRUE, selected variables available at each layer are used.

impute

boolean
If TRUE, mode or median based imputation is performed on the modality-specific predictions.

Returns

The current object is returned, with a meta training dataset assigned to the meta layer.

Method train()

Trains the current object. All leaners and the meta learner are trained.

Usage
Training$train(
  ind_subset = NULL,
  use_var_sel = FALSE,
  resampling_method = NULL,
  resampling_arg = list(),
  seed = NULL
)
Arguments
ind_subset

vector
ID subset to be used for training.

use_var_sel

boolean
If TRUE, variable selection is performed before training.

resampling_method

function
Function for internal validation. If not specify, the resampling function from the package caret is used for a 10-folds cross-validation.

resampling_arg

list
List of arguments to be passed to the function.

seed

integer
Random seed. Default is NULL, which generates the seed from R.

Returns

The current object is returned, with each learner trained on each layer.

Method predict()

Compute predictions for a testing object.

Usage
Training$predict(testing, ind_subset = NULL)
Arguments
testing

Testing
A new testing object to be predicted.

ind_subset

vector
Vector of IDs to be predicted.

Returns

The predicted object. All layers and the meta layer are predicted. This is the final predicted object.

Method varSelection()

Variable selection on the current training object.

Usage
Training$varSelection(ind_subset = NULL, verbose = TRUE)
Arguments
ind_subset

vector
ID subset of individuals to be used for variable selection.

verbose

boolean
Warning messages will be displayed if set to TRUE.

Returns

The current layer is returned with the resulting model.

Method getTargetValues()

Gather target values from all layer.

Usage
Training$getTargetValues()
Returns

A data.frame containing individuals IDs and corresponding target values.

Method getIndIDs()

Gather individual IDs from all layer.

Usage
Training$getIndIDs()
Returns

A data.frame containing individuals IDs.

Method getLayer()

Get a layer of a given ID.

Usage
Training$getLayer(id)
Arguments
id

character
The ID of the layer to be returned.

Returns

The TrainLayer object is returned for the given ID.

Method getTrainMetaLayer()

Getter of the meta layer.

Usage
Training$getTrainMetaLayer()
Returns

Object from class TrainMetaLayer

Method getModel()

Retrieve models from all layer.

Usage
Training$getModel()
Returns

A list containing all (base and meta) models.

Method getData()

Retrieve modality-specific predictions.

Usage
Training$getData()
Returns

A list containing all (base and meta) models.

Method removeLayer()

Remove a layer of a given ID.

Usage
Training$removeLayer(id)
Arguments
id

character
The ID of the layer to be removed.

Returns

The TrainLayer object is returned for the given ID.

Method removeTrainMetaLayer()

Remove the meta layer from the current Training object.

Usage
Training$removeTrainMetaLayer()

Method getIndCol()

Getter of the individual column name.

Usage
Training$getIndCol()

Method getTarget()

Getter of the target variable name.

Usage
Training$getTarget()

Method getVerbose()

Getter of the verbose setting.

Usage
Training$getVerbose()

Method getUseVarSel()

Getter of the use_var_sel field.

Usage
Training$getUseVarSel()

Method getVarSelDone()

Getter of the use_var_sel field.

Usage
Training$getVarSelDone()

Method increaseNbTrainedLayer()

Increase the number of trained layer.

Usage
Training$increaseNbTrainedLayer()

Method checkTargetExist()

Check whether a target object has already been stored.

Usage
Training$checkTargetExist()
Returns

Boolean value

Method getTargetObj()

Getter of the target object.

Usage
Training$getTargetObj()

Method getProblemTyp()

Getter of the problem type.

Usage
Training$getProblemTyp()

Method setImpute()

Set imputation action na.action.

Usage
Training$setImpute(impute)
Arguments
impute

character
How to handle missing values.

Method testOverlap()

Test that individuals overlap over layers. At least five individuals must overlapped.

Usage
Training$testOverlap()

Method upset()

UpSet plot to show an overview of the overlap of individuals across various layers.

Usage
Training$upset(...)
Arguments
...

any
Further parameters to be passed to the upset function from package UpSetR.

Method summary()

Generate training summary

Usage
Training$summary()

See Also

TrainLayer

Testing and Predicting

Varsel Class

Description

This class implements a learner. A VarSel object can only exist as a component of a TrainLayer or a TrainMetaLayer object.

Methods

Public methods

Method new()

Variable selection parameter list.

Learner ID.

Usage
VarSel$new(
  id,
  package = NULL,
  varsel_fct,
  varsel_param,
  train_layer,
  na_action = "na.rm"
)
Arguments
id

character
Package that implements the variable selection function. If NULL, the variable selection function is called from the current environment.

package

character
Variable selection function name. Note: Variable selection functions, except Boruta, must return a vector of selected variables.

varsel_fct

character
Variable selection parameters.

varsel_param

list
Layer on which the learner is stored.

train_layer

TrainLayer
The training layer where to store the learner.

na_action

character
Handling of missing values in meta-data. Set to "na.keep" to keep missing values, "na.rm" to remove individuals with missing values or "na.impute" (only applicable on meta-data) to impute missing values in meta-data. Only median and mode based imputations are actually handled. With the "na.keep" option, ensure that the provided learner can handle missing values. If TRUE, the individuals with missing predictor values will be removed from the training dataset.

Method print()

Printer

Usage
VarSel$print(...)
Arguments
...

any

Method summary()

Summary

Usage
VarSel$summary(...)
Arguments
...

any

Method interface()

Learner and prediction parameter interface. Use this function to provide how the following parameters are named in the learning function (lrn_fct) you provided when creating the learner, or in the predicting function.

Usage
VarSel$interface(
  x = "x",
  y = "y",
  object = "object",
  data = "data",
  extract_var_fct = NULL
)
Arguments
x

string
Name of the argument to pass the matrix of independent variables in the original learning function.

y

string
Name of the argument to pass the response variable in the original learning function.

object

string
Name of the argument to pass the model in the original predicting function.

data

character
Name of the argument to pass new data in the original predicting function.

extract_var_fct

character or function
If the variable selection function that is called does not return a vector, then use this argument to specify a (or a name of a) function that can be used to extract vector of selected variables. Default value is NULL, if selected variables are in a vector.

Method varSelection()

Tains the current learner (from class Lrner) on the current training data (from class TrainData).

Usage
VarSel$varSelection(ind_subset = NULL)
Arguments
ind_subset

vector
Individual ID subset on which the training will be performed.

Returns

The resulting model, from class Model, is returned.

Method getTrainLayer()

The current layer is returned.

Usage
VarSel$getTrainLayer()
Returns

TrainLayer object.

Method getId()

Getter of the current learner ID.

Usage
VarSel$getId()
Returns

The current learner ID.

Method getPackage()

Getter of the variable selection package implementing the variable selection function.

Usage
VarSel$getPackage()
Returns

The name of the package implementing the variable selection function.

Method getVarSubSet()

Getter of the list of selected variables.

Usage
VarSel$getVarSubSet()
Returns

List of selected variables..

Method getParamInterface()

The current parameter interface is returned.

Usage
VarSel$getParamInterface()
Returns

A data.frame of interface.

Method getNaAction()

The current layer is returned.

Usage
VarSel$getNaAction()

Method getExtractVar()

The function to extract selected variables is returned.

Usage
VarSel$getExtractVar()
Returns

A data.frame of interface.

The best layer-specific model is used as meta model.

Description

The meta learner is the best layer-specific learner. This function is intended to be (internally) used as meta-learner in fuseMLR.

Usage

bestLayerLearner(x, y, perf = NULL)

Arguments

x

data.frame
data.frame of predictors.

y

vector
True target observations. Either binary or two level factor variable.

perf

function
Function to compute layer-specific performance of learners. If NULL, the Brier Score (classification) or a mean squared error (regression) is used by default as performance measure. Otherwise, the performance function must accept two parameters: observed (observed values) and predicted (predicted values).

Value

A model object of class weightedMeanLeaner.

Examples

set.seed(20240624L)
x = data.frame(x1 = runif(n = 50L, min = 0, max = 1))
y = sample(x = 0L:1L, size = 50L, replace = TRUE)
my_best_model = bestLayerLearner(x = x, y = y)

Cobra Meta Learner

Description

The function cobra implements the COBRA (COmBined Regression Alternative), an aggregation method for combining predictions from multiple individual learners. This method aims to tune key parameters for achieving optimal predictions by averaging the target values of similar candidates in the training dataset's predictions. Only the training points that are sufficiently similar to the test point (based on the proximity threshold epsilon) are used for prediction. If no suitable training points are found, the function returns NA.

Usage

cobra(x, y, tune = "epsilon", k_folds = NULL, eps = NULL)

Arguments

x

data.frame
A training data, where rows are observations and columns are predictions from individual learners. Use NA for missing predictions.

y

vector
A vector containing the training targets. This can be a binary or two-level factor variable.

tune

character
A character value specifying the tuning mode:

  • "alpha_epsilon": Tunes both alpha (number of learners) and epsilon (proximity threshold) via cross-validation.

  • "epsilon": Tunes epsilon only via cross-validation.

  • "user": No tuning; the user provides an optimal epsilon. #' The default value is epsilon.

k_folds

integer
Number of folds for cross-validation when tune = "alpha_epsilon" or "epsilon". Default is 10.

eps

numeric
A numeric value for the proximity threshold, used only when tune = "user". Defaults to 0.1.

Value

An object of class cobra containing the training data, target values, and chosen parameters.

References

Biau, G., Fischer, A., Guedj, B., & Malley, J. D. (2014). COBRA: A combined regression strategy. The Journal of Multivariate Analysis 46:18-28

Examples

# Example usage
set.seed(123)
x_train <- data.frame(a = runif(10L), b = runif(10L))
y_train <- sample(0L:1L, size = 10L, replace = TRUE)

# Train the model with epsilon optimization
cobra_model <- cobra(x = x_train, y = y_train, tune = "epsilon", k_folds = 2)

# Make predictions on new data
set.seed(156)
x_new <- data.frame(a = runif(5L), b = runif(5L))
prediction <- predict(object = cobra_model, data = x_new)

Create COBRA Predictions

Description

The createCobraPred function calculates predictions by averaging the target values of all the nearest candidates in the training dataset. Only the training points that are within the specified proximity (eps) to the test point are used to determine the prediction. If no suitable training points are found, the function returns NA as the prediction.

Usage

createCobraPred(
  train,
  test,
  n_train,
  n_test,
  nlearners,
  eps,
  alpha,
  train_target
)

Arguments

train

A matrix representing the training data. Rows represent observations, and columns contain predictions from individual learners for these observations. In cases where a prediction is unavailable for a specific observation, NA is used.

test

A matrix representing the test data. Rows represent observations, and columns contain predictions from individual learners for these observations. In cases where a prediction is unavailable for a specific observation, NA is used.

n_train

An integer specifying the number of training observations.

n_test

An integer specifying the number of test observations.

nlearners

An integer representing the number of learners.

eps

A numeric value representing the threshold for proximity between two predictions.

alpha

A value that determines the optimal number of learners in the neighborhood (only for alpha optimization).

train_target

A vector containing the target values for the training dataset

Create Difference

Description

The createDif function computes the difference between the maximum and minimum predictions in a dataset.

Usage

createDif(x)

Arguments

x

Predictions vector

Create Loss

Description

Create Loss

Usage

createLoss(pred, target)

Arguments

pred

A vector of predictions.

target

A vector of target values.

createTestLayer

Description

Creates and stores a TestLayer on the Testing object passed as argument.

Usage

createTestLayer(testing, test_layer_id, test_data)

Arguments

testing

Testing
Testing object where the created layer will be stored.

test_layer_id

character
ID of the testing layer to be created.

test_data

data.frame
Data modality to be stored in TestData.

Value

The updated Testing object (with the new layer) is returned.

createTesting

Description

Creates a Testing object.

Usage

createTesting(id, ind_col, verbose = TRUE)

Arguments

id

character
Testing id.

ind_col

character
Name of column of individuals IDs in testing data.frame.

verbose

boolean
Warning messages will be displayed if set to TRUE.

Value

A Testing object.

createTrainLayer

Description

Creates and stores a TrainLayer on the Training object passed as argument. The main components of a training layer are training data modality, a variable selection methods, and a modality-specific learner.

Usage

createTrainLayer(
  training,
  train_layer_id,
  train_data,
  varsel_package = NULL,
  varsel_fct = NULL,
  varsel_param = list(),
  lrner_package = NULL,
  lrn_fct,
  param_train_list = list(),
  param_pred_list = list(),
  na_action = "na.rm",
  x_varsel = "x",
  y_varsel = "y",
  x_lrn = "x",
  y_lrn = "y",
  object = "object",
  data = "data",
  extract_pred_fct = NULL,
  extract_var_fct = NULL
)

Arguments

training

Training
Training object for storing the created layer.

train_layer_id

character
ID of the TrainLayer to be created.

train_data

data.frame
Data modality to be stored on the layer.

varsel_package

character
Package name containing the variable selection algorithm function. Defaults to NULL if the function exists in the current working environment.

varsel_fct

character
Variable selection function name. Default value is NULL for no variable selection. If specified, the function must accept at least two parameters: x (predictors) and y (response values), and return a vector of selected variables. Alternatively, use the interface parameters x_varsel and y_varsel to map the original argument names, and extract_var_fct to specify how to extract the vector of selected variables. An exception is made for the Boruta function, which includes an internal adjustment and requires no additional modifications.

varsel_param

list
List of arguments to be passed to varsel_fct.

lrner_package

character
Name of the package containing the learning algorithm function. Defaults to NULL if the function is available in the current working environment.

lrn_fct

character
Name of the learning function. The function must accept at least two parameters: x (predictors) and y (response values) and return a model. Alternatively, use the interface parameters x_lrn and y_lrn to map these names to the original arguments in your function. The returned model must support the generic predict function (with arguments object and data) to generate predictions for new data. Predictions should be either a vector or a list containing a vector named predictions with the predicted values.
If the arguments object and data have different names in your predict function, use the interface parameters below to map them to the original names. Additionally, if predictions are stored as a matrix or data.frame (e.g., predicted probabilities for dichotomous classification), only the second column (assumed to be class 1 probabilities) will be used. If the predicted values are not returned in one of the formats mentioned above, use the extract_pred_fct argument below to specify how to extract the predicted values from the prediction object.

param_train_list

character
List of arguments to be passed to lrn_fct.

param_pred_list

character
List of arguments to be passed to predict when generating predictions.

na_action

character
Handling of missing values in data during training. Set to "na.keep" to retain missing values, or "na.rm" to remove instances with missing values.

x_varsel

character
If the name of the argument used by the provided original variable selection function to pass the matrix of independent variable is not x, use this argument to specify how it is called in the provided function.

y_varsel

character
If the name of the argument used by the provided original variable selection function to pass the target variable is not y, use this argument to specify how it is called in the provided function.

x_lrn

character
If the name of the argument used by the provided original learning function to pass the matrix of independent variable is not x, use this argument to specify how it is called in the provided function.

y_lrn

character
If the name of the argument used by the provided original learning function to pass the target variable is not y, use this argument to specify how it is called in the provided function.

object

character
The generic function predict uses the parameter object to pass a model. If the corresponding argument is named differently in your predict function, specify its name.

data

character
The generic function predict uses a parameter data to pass new data. If the corresponding argument is named differently in your predict function, specify the name.

extract_pred_fct

⁠character or function⁠
If the predict function called for the model does not return a vector, use this argument to specify a function (or the name of a function) to extract the vector of predictions. The default value is NULL if predictions are returned as a vector.

extract_var_fct

⁠character or function⁠
If the variable selection function does not return a vector, use this argument to specify a function (or the name of a function) to extract the vector of selected variables.

Value

The updated Training object (with the new layer) is returned.

References

Fouodo C.J.K, Bleskina M. and Szymczak S. (2024). fuseMLR: An R package for integrative prediction modeling of multi-omics data, paper submitted.

See Also

createTrainMetaLayer and fusemlr.

createTrainMetaLayer

Description

Creates and store a TrainMetaLayer on the Training object passed as argument. The meta-layer encapsulates the meta-learner and the fold predictions (internally created) of the layer-specific base models.

Usage

createTrainMetaLayer(
  training,
  meta_layer_id,
  lrner_package = NULL,
  lrn_fct,
  param_train_list = list(),
  param_pred_list = list(),
  na_action = "na.impute",
  x_lrn = "x",
  y_lrn = "y",
  object = "object",
  data = "data",
  extract_pred_fct = NULL
)

Arguments

training

Training
Training object for storing the created meta-layer.

meta_layer_id

character
ID of the layer to be created.

lrner_package

character
Package name containing the variable selection algorithm function. Defaults to NULL if the function exists in the current working environment.

lrn_fct

character
Name of the learning function. The function must accept at least two parameters: x (predictors) and y (response values), and return a model. If not, use the interface parameters x_lrn and y_lrn below to map these argument names to the original arguments in your function. The returned model must support the generic predict function (with arguments object and data) to make predictions for new data, and the predictions should be a vector or a list containing a vector called predictions with the predicted values. If the arguments object and data are named differently in your predict function, use the interface parameters object and data below to specify the original names. See the details below about meta-learners.

param_train_list

character
List of arguments to be passed to lrn_fct.

param_pred_list

list
List of arguments to be passed to predict when computing predictions.

na_action

character
Handling of missing values in modality-specific predictions during training. Set to "na.keep" to keep missing values, "na.rm" to remove individuals with missing values or "na.impute" to impute missing values in modality-specific predictions. Only median and mode based imputations are actually handled. With the "na.keep" option, ensure that the provided meta-learner can handle missing values.

x_lrn

character
If the argument name used by the provided original function to pass the matrix of independent variables is not x, use this argument to specify the name used in the function.

y_lrn

character
If the argument name used by the provided original function to pass the target variable is not y, use this argument to specify the name used in the function.

object

character
The generic function predict uses a parameter object to pass a model. If the corresponding argument is named differently in your predict function, specify the name.

data

character
The generic function predict uses a parameter data to pass new data. If the corresponding argument is named differently in your predict function, specify the name.

extract_pred_fct

⁠character or function⁠
If the predict function that is called for the model does not return a vector, then use this argument to specify a (or a name of a) function that can be used to extract vector of predictions. Defaults to NULL, if predictions are a vector.

Details

Internal meta-learners are available in the package.

The cobra meta-learner implements the COBRA (COmBined Regression Alternative), an aggregation method for combining predictions from multiple individual learners (Biau et al. 2014). This method aims to tune key parameters for achieving optimal predictions by averaging the target values of similar candidates in the training dataset's predictions. Only the training points that are sufficiently similar to the test point (based on the proximity threshold epsilon) are used for prediction. If no suitable training points are found, the function returns NA.

The weightedMeanLearner evaluates the prediction performance of modality-specific learners and uses these estimates to weight the base models, aggregating their predictions accordingly.

The bestLayerLearner evaluates the prediction performance of modality-specific learners and returns predictions made by the best learner as the meta-prediction.

Beyond the internal meta-learners, any other learning algorithm can be used.

Value

The updated Training object (with the new layer) is returned.

References

Fouodo C.J.K, Bleskina M. and Szymczak S. (2024). fuseMLR: An R package for integrative prediction modeling of multi-omics data, paper submitted.
Biau, G., Fischer, A., Guedj, B., & Malley, J. D. (2014). COBRA: A combined regression strategy. The Journal of Multivariate Analysis 46:18-28

See Also

createTrainLayer, varSelection, and fusemlr.

createTraining

Description

Creates a Training object. A training object is designed to encapsulate training layers and training meta-layer. Functions createTrainLayer and createTrainMetaLayer are available to add the training layer and the training meta-layer to a training object.

Usage

createTraining(
  id,
  target_df,
  ind_col,
  target,
  problem_type = "classification",
  verbose = TRUE
)

Arguments

id

character
Training's ID.

target_df

data.frame
Observed target values. A data frame with two columns: individual IDs and response variable values.

ind_col

character
Name of column of individuals IDs.

target

character
Name of the target variable.

problem_type

character
Either "classification" or "regression".

verbose

boolean
Warning and processing information (including those of cross-validation) will be displayed if set to TRUE.

Value

The created Training object is returned.

See Also

createTrainLayer, createTrainMetaLayer and fusemlr.

Create weights for COBRA Predictions

Description

The createWeights function is used to calculate weights for predictions.

Usage

createWeights(train, test, n_train, n_test, nlearners, eps, alpha)

Arguments

train

A matrix representing the training data. Rows represent observations, and columns contain predictions from individual learners for these observations. In cases where a prediction is unavailable for a specific observation, NA is used.

test

A matrix representing the test data. Rows represent observations, and columns contain predictions from individual learners for these observations. In cases where a prediction is unavailable for a specific observation, NA is used.

n_train

An integer specifying the number of training observations.

n_test

An integer specifying the number of test observations.

nlearners

An integer representing the number of learners.

eps

A numeric value representing the threshold for proximity between two predictions.

alpha

A value that determines the optimal number of learners in the neighborhood (only for alpha optimization).

extractData

Description

Extracts data stored on each layers; base data and modality-specific predictions (for Training) are extracted.

Usage

extractData(object)

Arguments

object

⁠Training or Testing⁠
The object of interest.

Value

A list of data is returned.

extractModel

Description

Extracts models stored on each layers; base and meta models are extracted.

Usage

extractModel(training)

Arguments

training

Training
The Training object of interest.

Value

A list of models is returned.

fusemlr

Description

Trains the Training object passed as argument. A training object must contain the training layers and a training meta-layer. A training layer encapsulates data modalities, a variable selection method and a learner. Use the function createTraining to create a training object, createTrainLayer to add training layers to the created training object, and createTrainMetaLayer to add a meta-layer with the corresponding meta-learner to the training object. The function fusemlr is designed to train all training layers and the meta-learner. After training the layer-specific base models and the meta-model will be stored in the training object which can be used for predictions.

Usage

fusemlr(
  training,
  ind_subset = NULL,
  use_var_sel = FALSE,
  resampling_method = NULL,
  resampling_arg = list(),
  seed = NULL
)

Arguments

training

Training
Training object for storing training layers.

ind_subset

vector
ID subset to be used for training.

use_var_sel

boolean
If TRUE and no variable selection has been performed for the provide training object, then a variable selection will proceed the training. Otherwise, if variable selection has been previously performed, the selected variables will be used for training.

resampling_method

function
Function for internal validation. If not specify, the resampling function from the package caret is used for a 10-folds cross-validation.

resampling_arg

list
List of arguments to be passed to the function.

seed

integer
Random seed used for resampling. Default is NULL, which generates the seed from R.

Value

The current object is returned, with each learner trained on each layer.

References

Fouodo C.J.K, Bleskina M. and Szymczak S. (2024). fuseMLR: An R package for integrative prediction modeling of multi-omics data, paper submitted.

See Also

createTrainLayer, createTrainMetaLayer, extractModel and extractData.

Simulated multiomics data for 70 training participants and 23 testing participants, each with an effect size of 20 on each layer. Each layer includes 50 participants for training and 20 for testing. Participants do not perfectly overlap across layers. The simulation is based on the R package interSIM.

Description

The dataset is a list containing training and testing data, called training and testing respectively. Each data is a list containing the following multi_omics at each layer.

Usage

data(multi_omics)

Format

A list with training and testing data contaning methylation, gene expressions and protein expressions data.

Details

predict.Training

Description

Computes predictions for the Testing object passed as argument.

Usage

## S3 method for class 'Training'
predict(object, testing, ind_subset = NULL, ...)

Arguments

object

Training
A trained Training object to be used to compute predictions.

testing

Testing
A new testing object to be predicted.

ind_subset

vector
Vector of IDs to be predicted.

...

any
Further arguments passed to or from other methods.

Value

The final predicted object. All layers and the meta layer are predicted.

Best specific Learner prediction.

Description

Predict function for models from class bestLayerLearner.

Usage

## S3 method for class 'bestLayerLearner'
predict(object, data, ...)

Arguments

object

bestLayerLearner
An object from class bestLayerLearner

data

data.frame
New data to predicted.

...

any
Further arguments passed to or from other methods.

Value

Predicted target values are returned.

Examples

set.seed(20240625)
x = data.frame(x1 = runif(n = 50L, min = 0, max = 1))
y <- sample(x = 0:1, size = 50L, replace = TRUE)
my_model <- bestLayerLearner(x = x, y = y)
x_new <- data.frame(x1 = rnorm(10L))
my_predictions <- predict(object = my_model, data = x_new)

Predict Using COBRA object

Description

#' The predict.cobra function makes predictions on new data using a trained COBRA object.

Usage

## S3 method for class 'cobra'
predict(object, data, ...)

Arguments

object

An object of class "cobra" created by the cobra function.

data

A data.frame of new data, where rows are observations and columns are predictions from individual learners. Use NA for missing predictions.

...

Additional arguments (currently not used).

Value

A vector of predictions for the new data.

Examples

# Example usage
set.seed(123)
x_train <- data.frame(a = rnorm(10L), b = rnorm(10L))
y_train <- sample(0L:1L, size = 10L, replace = TRUE)

# Train the model with epsilon optimization
cobra_model <- cobra(x = x_train, y = y_train, tune = "epsilon")

# Make predictions on new data
set.seed(156)
x_new <- data.frame(a = rnorm(5L), b = rnorm(5L))
prediction <- predict(object = cobra_model, data = x_new)

Weighted mean prediction.

Description

Predict function for models from class weightedMeanLearner.

Usage

## S3 method for class 'weightedMeanLearner'
predict(object, data, na_rm = FALSE, ...)

Arguments

object

weightedMeanLearner(1)
An object from class weightedMeanLearner

data

data.frame
data.frame to be predicted.

na_rm

boolean
Removes NAs when TRUE.

...

any
Further arguments.

Value

Predicted target values are returned.

Examples

set.seed(20240625)
x <- data.frame(x1 = rnorm(50L))
y <- sample(x = 0:1, size = 50L, replace = TRUE)
my_model <- weightedMeanLearner(x = x, y = y)
x_new <- data.frame(x1 = rnorm(10L))
my_predictions <- predict(object = my_model, data = x_new)

Testing object Summaries

Description

Summaries a fuseMLR Testing object.

Usage

## S3 method for class 'Testing'
summary(object, ...)

Arguments

object

Testing
The Testing object of interest.

...

any
Further arguments.

Training object Summaries

Description

Summaries a fuseMLR Training object.

Usage

## S3 method for class 'Training'
summary(object, ...)

Arguments

object

Training
The Training object of interest.

...

any
Further arguments.

upsetplot

Description

An upset plot of overlapping individuals.

Usage

upsetplot(object, ...)

Arguments

object

⁠Training or Testing⁠
Training or testing object for each the upset plot will be created.

...

any
Further arguments to be passed to the upset function from package UpSetR.

varSelection

Description

Variable selection on the training object passed as argument.

Usage

varSelection(training, ind_subset = NULL)

Arguments

training

Training
Training object for storing the created layer.

ind_subset

vector
ID subset of individuals to be used for variable selection.

Value

A data.frame with two columns: layer and selected variables.

References

Fouodo C.J.K, Bleskina M. and Szymczak (2024). fuseMLR: An R package for integrative prediction modeling of multi-omics data, paper submitted.

The weighted mean meta-learner

Description

Modality-specific learner are assessed and weighted based on their predictions. This function is intended to be (internally) used as meta-learner in fuseMLR.

Usage

weightedMeanLearner(x, y, weighted = TRUE, perf = NULL, na_rm = FALSE)

Arguments

x

data.frame
Modality-specific predictions. Each column of the data.frame content the predictions a specific learner.

y

vector
True target values. If classification, either binary or two level factor variable.

weighted

boolean
If TRUE, a weighted sum is computed. As default, weights are estimated based on Brier Score for classification setting and mean squared error for regression. Otherwise, use argument perf below to specify the function to use estimate learner performance.

perf

function
Function to compute layer-specific performance of learners. If NULL, the Brier Score (classification) or a mean squared error (regression) is used by default as performance measure. Otherwise, the performance function must accept two parameters: observed (observed values) and predicted (predicted values).

na_rm

boolean
Should missing values be removed when computing the weights?

Value

Object of class weightedMeanLearner with the vector of estimated weights pro layer.

Examples

set.seed(20240624L)
x = data.frame(x1 = runif(n = 50L, min = 0, max = 1),
               x2 = runif(n = 50L, min = 0, max = 1))
y = sample(x = 0L:1L, size = 50L, replace = TRUE)
my_model = weightedMeanLearner(x = x, y = y)