Package 'pathMED' reference manual

Title:	Scoring Personalized Molecular Portraits
Description:	PathMED is a collection of tools to facilitate precision medicine studies with omics data (e.g. transcriptomics). Among its funcionalities, genesets scores for individual samples may be calculated with several methods. These scores may be used to train machine learning models and to predict clinical features on new data. For this, several machine learning methods are evaluated in order to select the best method based on internal validation and to tune the hyperparameters. Performance metrics and a ready-to-use model to predict the outcomes for new patients are returned.
Authors:	Jordi Martorell-Marugán [cre, aut] (ORCID: <https://orcid.org/0000-0002-5186-0735>), Daniel Toro-Domínguez [aut] (ORCID: <https://orcid.org/0000-0001-8440-312X>), Raúl López-Domínguez [aut] (ORCID: <https://orcid.org/0000-0001-8634-117X>), Iván Ellson [aut] (ORCID: <https://orcid.org/0000-0001-6307-3141>)
Maintainer:	Jordi Martorell-Marugán <[email protected]>
License:	GPL-2
Version:	1.5.1
Built:	2026-06-03 18:30:51 UTC
Source:	https://github.com/bioc/pathMED

Annotate the pathways from a scores matrix

Description

Annotate the pathways from a scores matrix

Usage

ann2term(scoresMatrix)
ann2term(scoresMatrix)

Arguments

scoresMatrix

Matrix with pathways IDs as row names

Value

A data frame with the input IDs and their corresponding terms

Author(s)

Raúl López-Domínguez, [email protected]

Jordi Martorell-Marugán, [email protected]

Daniel Toro-Dominguez, [email protected]

References

Toro-Domínguez, D. et al (2022). Scoring personalized molecular portraits identify Systemic Lupus Erythematosus subtypes and predict individualized drug responses, symptomatology and disease progression . Briefings in Bioinformatics. 23(5)

Examples

data(pathMEDExampleData)
scoresExample <- getScores(pathMEDExampleData, geneSets = "tmod", 
                             method = "GSVA")
annotatedTerms <- ann2term(scoresExample)
data(pathMEDExampleData)
scoresExample <- getScores(pathMEDExampleData, geneSets = "tmod", 
                             method = "GSVA")
annotatedTerms <- ann2term(scoresExample)

Create a reference data object for input to the pathMED functions

Description

Create a reference data object for input to the pathMED functions

Usage

buildRefObject(data, metadata = NULL, groupVar, controlGroup, use.assay = 1)
buildRefObject(data, metadata = NULL, groupVar, controlGroup, use.assay = 1)

Arguments

data

A list of matrices, data frames, ExpressionSets or SummarizedExperiments with samples in columns and features in rows. A single matrix, dataframe, ExpressionSet or SummarizedExperiment may be also used.

metadata

A list of data frames or a single data frame with information for each sample. Samples in rows and variables in columns. If a list of ExpressionSets or SummarizedExperiments are used as @data, it is not necessary to provide @metadata.

groupVar

Character or list of characters indicating the column name of @metadata classifying the samples in controls and cases. If several metadata objects are provided a @groupVar can be specified for each metadata.

controlGroup

Character or list of characters indicating which @groupVar level corresponds to the control group, usually healthy samples. All other samples will be considered as cases, usually disease samples. If several @groupVar are provided a @controlGroup can be specified for each @groupVar

use.assay

If SummarizedExperiments are used, the number of the assay to extract the data.

Value

A refObject that serves as input for mScores_createReference and dissectDB functions.

Author(s)

Iván Ellson, [email protected]

Jordi Martorell-Marugán, [email protected]

Daniel Toro-Dominguez, [email protected]

References

Examples

data(refData)

refObject <- buildRefObject(
    data = list(
        refData$dataset1, refData$dataset2,
        refData$dataset3, refData$dataset4
    ),
    metadata = list(
        refData$metadata1, refData$metadata2,
        refData$metadata3, refData$metadata4
    ),
    groupVar = "group",
    controlGroup = "Healthy_sample"
)

## Also works with a metadata for all datasets
metadata <- rbind(
    refData$metadata1, refData$metadata2,
    refData$metadata3, refData$metadata4
)
refObject <- buildRefObject(
    data = list(
        refData$dataset1, refData$dataset2,
        refData$dataset3, refData$dataset4
    ),
    metadata = metadata,
    groupVar = "group",
    controlGroup = "Healthy_sample"
)

data(refData)

refObject <- buildRefObject(
    data = list(
        refData$dataset1, refData$dataset2,
        refData$dataset3, refData$dataset4
    ),
    metadata = list(
        refData$metadata1, refData$metadata2,
        refData$metadata3, refData$metadata4
    ),
    groupVar = "group",
    controlGroup = "Healthy_sample"
)

## Also works with a metadata for all datasets
metadata <- rbind(
    refData$metadata1, refData$metadata2,
    refData$metadata3, refData$metadata4
)
refObject <- buildRefObject(
    data = list(
        refData$dataset1, refData$dataset2,
        refData$dataset3, refData$dataset4
    ),
    metadata = metadata,
    groupVar = "group",
    controlGroup = "Healthy_sample"
)

Split pathways into coexpressed subpathways

Description

Split pathways into coexpressed subpathways

Usage

dissectDB(
  refObject,
  geneSets,
  minPathSize = 10,
  minSplitSize = 3,
  maxSplits = NULL,
  explainedVariance = 60,
  percSharedGenes = 90,
  use.assay = 1
)
dissectDB(
  refObject,
  geneSets,
  minPathSize = 10,
  minSplitSize = 3,
  maxSplits = NULL,
  explainedVariance = 60,
  percSharedGenes = 90,
  use.assay = 1
)

Arguments

refObject

A refObject object structure: a list of lists, each one with a cases omic matrix and controls omic matrix (named as Disease and Healthy). It can be constructed with the buildRefObject function. A list with one or more expression matrices, ExpressionSets or SummarizedExperiments without controls, can also be used. Data should be normalized and log2-transformed. Feature names must match the gene sets nomenclature. To use preloaded databases, they must be gene symbols.

geneSets

A named list with each gene set, or the name of one preloaded database (go_bp, go_cc, go_mf, kegg, reactome, pharmgkb, lincs, ctd, disgenet, hpo, wikipathways, tmod) or a GeneSetCollection.

minPathSize

numeric, minimum number of genes in a pathway to consider splitting it.

minSplitSize

numeric, minimum number of genes in a subpathway. Smaller splits will be merged with the closest coexpressed subpathway.

maxSplits

numeric, maximum number of subpathways for a pathway. If NULL (default), there is not limit.

explainedVariance

numeric, percentage of cumulative variance explained within a pathway. This parameter is used to select the number of subdivisions of a pathway that manage to explain at least the percentage of variance defined by explainedVariance.

percSharedGenes

numeric, minimum percentage of common genes across datasets to merge them before clustering. If NULL or this percentage is not reached, clustering is performed for each dataset independently and consensus subpathways are obtained from co-occurrence across datasets.

use.assay

If SummarizedExperiments are used, the number of the assay to extract the data.

Value

A list with the subpathways.

Author(s)

Jordi Martorell-Marugán, [email protected]

Daniel Toro-Dominguez, [email protected]

References

Examples

data(refData)

refObject <- buildRefObject(
    data = list(
        refData$dataset1, refData$dataset2,
        refData$dataset3, refData$dataset4
    ),
    metadata = list(
        refData$metadata1, refData$metadata2,
        refData$metadata3, refData$metadata4
    ),
    groupVar = "group",
    controlGroup = "Healthy_sample"
)

set.seed(123)
custom.tmod <- dissectDB(refObject, geneSets = "tmod")
data(refData)

refObject <- buildRefObject(
    data = list(
        refData$dataset1, refData$dataset2,
        refData$dataset3, refData$dataset4
    ),
    metadata = list(
        refData$metadata1, refData$metadata2,
        refData$metadata3, refData$metadata4
    ),
    groupVar = "group",
    controlGroup = "Healthy_sample"
)

set.seed(123)
custom.tmod <- dissectDB(refObject, geneSets = "tmod")

Preloaded gene sets

Description

genesetsData was constructed from the GeneCodis database (https://genecodis.genyo.es/)

Usage

data(genesetsData)
data(genesetsData)

Format

An object of class "list" with one list per database. Each database consists on a list of gene sets, containing the gene symbols associated to it.

Calculate pathways scores for a dataset

Description

Calculate pathways scores for a dataset

Usage

getScores(
  inputData,
  geneSets,
  method = "GSVA",
  labels = NULL,
  cores = 1,
  use.assay = 1,
  ...
)
getScores(
  inputData,
  geneSets,
  method = "GSVA",
  labels = NULL,
  cores = 1,
  use.assay = 1,
  ...
)

Arguments

inputData

Matrix, data frame, ExpressionSet or SummarizedExperiment with omics data. Feature names must match the gene sets nomenclature. To use preloaded databases, they must be gene symbols.

geneSets

A named list with each gene set, or the name of one preloaded database (go_bp, go_cc, go_mf, kegg, reactome, pharmgkb, lincs, ctd, disgenet, hpo, wikipathways, tmod) or a GeneSetCollection. For using network methods, a data frame including columns: "source","target","weight" and "mor" (optional).

method

Scoring method: M-Scores, GSVA, ssGSEA, singscore, Plage, Z-score, AUCell, MDT, MLM, ORA, UDT, ULM, FGSEA, norm_FGSEA, WMEAN, norm_WMEAN, corr_WMEAN, WSUM, norm_WSUM or corr_WSUM.

labels

(Only for M-Scores) Vector with the samples class labels (0 or "Healthy" for control samples). Optional.

cores

Number of cores to be used.

use.assay

If SummarizedExperiments are used, the number of the assay to extract the data.

...

Additional parameters for the scoring functions.

Value

A list with the results of each of the analyzed regions. For each region type, a data frame with the results and a list with the probes associated to each region are generated. In addition, this list also contains the input methData, pheno and platform objects

Author(s)

Jordi Martorell-Marugán, [email protected]

Daniel Toro-Dominguez, [email protected]

References

Examples

data(pathMEDExampleData)
scoresExample <- getScores(pathMEDExampleData, geneSets = "tmod", 
                             method = "GSVA")
data(pathMEDExampleData)
scoresExample <- getScores(pathMEDExampleData, geneSets = "tmod", 
                             method = "GSVA")

Prepare the models parameter for the trainModel function

Description

Prepare the models parameter for the trainModel function

Usage

methodsML(algorithms = c("rf", "knn", "nb"), outcomeClass, tuneLength = 20)
methodsML(algorithms = c("rf", "knn", "nb"), outcomeClass, tuneLength = 20)

Arguments

algorithms

Vector with one or more of these methods: 'glm', 'lm', 'lda', 'xgbTree', 'rf', 'knn', 'svmLinear', 'nnet', 'svmRadial', 'nb', 'lars','rpart', 'gamboost', 'ada', 'brnn', 'enet', or 'all' to use all algorithms

outcomeClass

Predicted variable type ('character' or 'numeric')

tuneLength

maximum number of tuning parameter combinations

Value

A list with the selected models ready to use as the 'models' parameter in the trainModel function

Author(s)

Jordi Martorell-Marugán, [email protected]

Daniel Toro-Dominguez, [email protected]

References

Examples

models <- methodsML(c("rf", "knn"), tuneLength = 20,
                    outcomeClass = "character")
models <- methodsML(c("rf", "knn"), tuneLength = 20,
                    outcomeClass = "character")

Create a reference dataset based on M-scores

Description

Create a reference dataset based on M-scores

Usage

mScores_createReference(refObject, geneSets, cores = 1)
mScores_createReference(refObject, geneSets, cores = 1)

Arguments

refObject

geneSets

A named list with each gene set, or the name of one preloaded database (go_bp, go_cc, go_mf, kegg, reactome, pharmgkb, lincs, ctd, disgenet, hpo, wikipathways, tmod) or a GeneSetCollection.

cores

Number of cores to be used.

Value

A list with three elements. The first one is a list with the M-scores for each dataset. The second one is the geneSet used for the analysis and the third one is the input data.

Author(s)

Jordi Martorell-Marugán, [email protected]

Daniel Toro-Dominguez, [email protected]

References

Examples

data(refData)

refObject <- buildRefObject(
    data = list(
        refData$dataset1, refData$dataset2,
        refData$dataset3, refData$dataset4
    ),
    metadata = list(
        refData$metadata1, refData$metadata2,
        refData$metadata3, refData$metadata4
    ),
    groupVar = "group",
    controlGroup = "Healthy_sample"
)

refMscore <- mScores_createReference(refObject, geneSets = "tmod")
data(refData)

refObject <- buildRefObject(
    data = list(
        refData$dataset1, refData$dataset2,
        refData$dataset3, refData$dataset4
    ),
    metadata = list(
        refData$metadata1, refData$metadata2,
        refData$metadata3, refData$metadata4
    ),
    groupVar = "group",
    controlGroup = "Healthy_sample"
)

refMscore <- mScores_createReference(refObject, geneSets = "tmod")

Filter pathways from the reference M-scores dataset

Description

Filter pathways from the reference M-scores dataset

Usage

mScores_filterPaths(
  MRef,
  min_datasets = round(length(MRef[[1]]) * 0.34),
  perc_samples = 10,
  Pcutoff = 0.05,
  plotMetrics = TRUE
)
mScores_filterPaths(
  MRef,
  min_datasets = round(length(MRef[[1]]) * 0.34),
  perc_samples = 10,
  Pcutoff = 0.05,
  plotMetrics = TRUE
)

Arguments

MRef

output from the mScores_createReference function

min_datasets

number of datasets that each pathway must meet the perc_samples threshold

perc_samples

minimun percentage of samples in a dataset in which a pathway must be significant

Pcutoff

P-value cutoff for significance

plotMetrics

Plot number of significant pathways selected based on the different combination of perc_samples and min_datasets parameters

Value

A list with the selected pathways

Author(s)

Jordi Martorell-Marugán, [email protected]

Daniel Toro-Dominguez, [email protected]

References

Examples

data(refData)

refObject <- buildRefObject(
    data = list(
        refData$dataset1, refData$dataset2,
        refData$dataset3, refData$dataset4
    ),
    metadata = list(
        refData$metadata1, refData$metadata2,
        refData$metadata3, refData$metadata4
    ),
    groupVar = "group",
    controlGroup = "Healthy_sample"
)

exampleRefMScore <- mScores_createReference(refObject, geneSets = "tmod")
relevantPaths <- mScores_filterPaths(exampleRefMScore, min_datasets = 3)
data(refData)

refObject <- buildRefObject(
    data = list(
        refData$dataset1, refData$dataset2,
        refData$dataset3, refData$dataset4
    ),
    metadata = list(
        refData$metadata1, refData$metadata2,
        refData$metadata3, refData$metadata4
    ),
    groupVar = "group",
    controlGroup = "Healthy_sample"
)

exampleRefMScore <- mScores_createReference(refObject, geneSets = "tmod")
relevantPaths <- mScores_filterPaths(exampleRefMScore, min_datasets = 3)

Estimate M-scores for a dataset without healthy controls

Description

Estimate M-scores for a dataset without healthy controls

Usage

mScores_imputeFromReference(
  inputData,
  geneSets,
  externalReference,
  nk = 5,
  distance.threshold = 30,
  cores = 1,
  use.assay = 1
)
mScores_imputeFromReference(
  inputData,
  geneSets,
  externalReference,
  nk = 5,
  distance.threshold = 30,
  cores = 1,
  use.assay = 1
)

Arguments

inputData

Data matrix, data frame ExpressionSet or SummarizedExperiment. Feature names must match the gene sets nomenclature. To use preloaded databases, they must be gene symbols.

geneSets

A named list with each gene set, or the name of one preloaded database (go_bp, go_cc, go_mf, kegg, reactome, pharmgkb, lincs, ctd, disgenet, hpo, wikipathways, tmod) or a GeneSetCollection.

externalReference

External reference created with the mScores_createReference function.

nk

Number of most similar samples from the external reference to impute M-scores.

distance.threshold

Only samples that do not surpass the mean Euclidean distance of distance.threshold (by default = 30) with the external reference are imputed. If NULL,impute all samples.

cores

Number of cores to be used.

use.assay

If SummarizedExperiments are used, the number of the assay to extract the data.

Value

Author(s)

Jordi Martorell-Marugán, [email protected]

Daniel Toro-Dominguez, [email protected]

References

Examples

data(refData, pathMEDExampleData)

refObject <- buildRefObject(
    data = list(
        refData$dataset1, refData$dataset2,
        refData$dataset3, refData$dataset4
    ),
    metadata = list(
        refData$metadata1, refData$metadata2,
        refData$metadata3, refData$metadata4
    ),
    groupVar = "group",
    controlGroup = "Healthy_sample"
)

refMScores <- mScores_createReference(refObject,
    geneSets = "tmod", cores = 1
)

exampleMScores <- mScores_imputeFromReference(pathMEDExampleData,
    geneSets = "tmod",
    externalReference = refMScores,
    distance.threshold = 50
)
data(refData, pathMEDExampleData)

refObject <- buildRefObject(
    data = list(
        refData$dataset1, refData$dataset2,
        refData$dataset3, refData$dataset4
    ),
    metadata = list(
        refData$metadata1, refData$metadata2,
        refData$metadata3, refData$metadata4
    ),
    groupVar = "group",
    controlGroup = "Healthy_sample"
)

refMScores <- mScores_createReference(refObject,
    geneSets = "tmod", cores = 1
)

exampleMScores <- mScores_imputeFromReference(pathMEDExampleData,
    geneSets = "tmod",
    externalReference = refMScores,
    distance.threshold = 50
)

Example of test gene expression data

Description

pathMEDExampleData was obtained from a dataset downloaded from NCBI GEO (GSE224705), that contains lupus patients treated with Micophenolate mofetil. The same preprocessing was done as for the datasets used to create refData. 40 patients were randomly selected, 20 samples from responding patients and 20 from non-responders.

Usage

data(pathMEDExampleData)
data(pathMEDExampleData)

Format

An object of class "data.frame" with genes in rows and samples in columns.

Metadata of test gene expression data

Description

Metadata from the dataset GSE224705. Response column conteins the information about the response and non-response to the drug for each sample.

Usage

data(pathMEDExampleMetadata)
data(pathMEDExampleMetadata)

Format

An object of class "data.frame" with samples in rows and variables in columns.

Predict conditions in external datasets

Description

Predict conditions in external datasets

Usage

predictExternal(
  testData,
  model,
  realValues = NULL,
  positiveClass = NULL,
  use.assay = 1
)
predictExternal(
  testData,
  model,
  realValues = NULL,
  positiveClass = NULL,
  use.assay = 1
)

Arguments

testData

Numerical matrix or data frame with the same features used for the model construction in rows, and the samples (new observations) in columns. An ExpressionSet may or SummarizedExperiment may also be used.

model

trainModel output or a caret-like model object

realValues

Optional, named vector (for numerical variables) or named factor (for categorical variables) with real values for each sample

positiveClass

Optional, positive class to get confusion matrix. Only needed when realValues = TRUE and for categorical variables

use.assay

If SummarizedExperiments are used, the number of the assay to extract the data.

Value

A dataframe with predictions (if realValues is not provided) or a list with the dataframe with predictions and a dataframe with the performance metrics (if realValues is provided)

Author(s)

Iván Ellson, [email protected]

Jordi Martorell-Marugán, [email protected]

Daniel Toro-Dominguez, [email protected]

References

Examples

data(refData)

commonGenes <- intersect(rownames(refData$dataset1),
                         rownames(refData$dataset2))
dataset1 <- refData$dataset1[commonGenes, ]
dataset2 <- refData$dataset2[commonGenes, ]

scoresExample <- getScores(dataset1, geneSets = "tmod", method = "Z-score")

set.seed(123)
trainedModel <- trainModel(
    inputData = scoresExample,
    metadata = refData$metadata1,
    var2predict = "group",
    models = methodsML("svmLinear",
        outcomeClass = "character"
    ),
    Koutter = 2,
    Kinner = 2,
    repeatsCV = 1
)

externalScores <- getScores(dataset2, geneSets = "tmod", method = "Z-score")
realValues <- refData$metadata2$group
names(realValues) <- rownames(refData$metadata2)
predictions <- predictExternal(externalScores, trainedModel,
    realValues = realValues
)

print(predictions)

data(refData)

commonGenes <- intersect(rownames(refData$dataset1),
                         rownames(refData$dataset2))
dataset1 <- refData$dataset1[commonGenes, ]
dataset2 <- refData$dataset2[commonGenes, ]

scoresExample <- getScores(dataset1, geneSets = "tmod", method = "Z-score")

set.seed(123)
trainedModel <- trainModel(
    inputData = scoresExample,
    metadata = refData$metadata1,
    var2predict = "group",
    models = methodsML("svmLinear",
        outcomeClass = "character"
    ),
    Koutter = 2,
    Kinner = 2,
    repeatsCV = 1
)

externalScores <- getScores(dataset2, geneSets = "tmod", method = "Z-score")
realValues <- refData$metadata2$group
names(realValues) <- rownames(refData$metadata2)
predictions <- predictExternal(externalScores, trainedModel,
    realValues = realValues
)

print(predictions)

Example of reference gene expression datasets

Description

refData contains processed gene expression data from four datasets, including Systemic Lupus Erythematosus patients and healthy controls. Raw data for each dataset were downloaded from NCBI GEO (GSE65391, GSE45291, GSE61635, and GSE72509, respectively). Platform-dependent preprocessing was performed following established guidelines (Martorell-Marugán et al., 2021). Gene expression data were log2-transformed, and probe sets were annotated to gene symbols. To reduce computational cost in examples, 20 patient and 10 control samples were randomly selected from each dataset.

Usage

data(refData)
data(refData)

Format

An object of class "list" containing eight objects (dataset1-4 and metadata1-4). Each dataset is a matrix of normalized gene expression values (genes in rows, samples in columns). Each metadata is a dataframe with two columns: samples and group.

Train ML models and perform internal validation

Description

Train ML models and perform internal validation

Usage

trainModel(
  inputData,
  metadata = NULL,
  models = methodsML(outcomeClass = "character"),
  var2predict,
  positiveClass = NULL,
  pairingColumn = NULL,
  Koutter = 5,
  Kinner = 4,
  repeatsCV = 5,
  priorStatDiscrete = "mcc",
  priorStatContinuous = "r",
  filterFeatures = NULL,
  filterSizes = seq(2, 100, by = 2),
  rerank = FALSE,
  continue_on_fail = TRUE,
  saveLogFile = NULL,
  modelEnsemble = FALSE,
  use.assay = 1
)
trainModel(
  inputData,
  metadata = NULL,
  models = methodsML(outcomeClass = "character"),
  var2predict,
  positiveClass = NULL,
  pairingColumn = NULL,
  Koutter = 5,
  Kinner = 4,
  repeatsCV = 5,
  priorStatDiscrete = "mcc",
  priorStatContinuous = "r",
  filterFeatures = NULL,
  filterSizes = seq(2, 100, by = 2),
  rerank = FALSE,
  continue_on_fail = TRUE,
  saveLogFile = NULL,
  modelEnsemble = FALSE,
  use.assay = 1
)

Arguments

inputData

Numerical matrix or data frame with samples in columns and features in rows. An ExpressionSet or SummarizedExperiment may also be used.

metadata

Data frame with information for each sample. Samples in rows and variables in columns. If @inputData is an ExpressionSet or SummarizedExperiment, the metadata will be extracted from it.

models

Named list with the ML models generated with caret::caretModelSpec function. methodsML function may be used to prepare this list.

var2predict

Character with the column name of the @metadata to predict

positiveClass

Value that must be considered as positive class (only for categoric variables). If NULL, the last class by alphabetical order is considered as the positive class.

pairingColumn

Optional. Character with the column name of the @metadata with pairing information (e.g. technical replicates). Paired samples will always be assigned to the same set (training/test) to avoid data leakage.

Koutter

Number of outter cross-validation folds. A list of integer with elements for each resampling iteration is admitted. Each list element is a vector of integers corresponding to the rows used for training on that iteration.

Kinner

Number of innter cross-validation folds (for parameter tuning).

repeatsCV

Number of repetitions of the parameter tuning process.

priorStatDiscrete

Performance metric used to select the top ML algorithm in classification tasks. One of the following ones: mcc, balacc, accuracy, recall, specificity, npv, precision, fscore.

priorStatContinuous

Performance metric used to select the top ML algorithm in regression tasks. One of the following ones: r, r2, RMSE, MAE, RMAE, RSE.

filterFeatures

"rfe" (Recursive Feature Elimination), "sbf" (Selection By Filtering) or NULL (no feature selection).

filterSizes

Only for filterFeatures = "rfe". A numeric vector of integers corresponding to the number of features that should be retained.

rerank

Only for filterFeatures = "rfe". A boolean indicating if the variable importance must be re-calculated each time features are removed.

continue_on_fail

Whether or not to continue training the models if any of them fail.

saveLogFile

Path to a .txt file in which to save error and warning messages.

modelEnsemble

Logical. If TRUE, evaluates an additional stacked ensemble that combines predictions from the valid trained algorithms.

use.assay

If SummarizedExperiments are used, the number of the assay to extract the data.

Value

A list with four elements. The first one is the model. The second one is a table with different metrics obtained. The third one is a list with the best parameters selected in tuning process. The last element contains data for AUC plots

Author(s)

Jordi Martorell-Marugán, [email protected]

Daniel Toro-Dominguez, [email protected]

References

Examples

data(pathMEDExampleData, pathMEDExampleMetadata)

scoresExample <- getScores(pathMEDExampleData, geneSets = "tmod", 
                             method = "GSVA")

modelsList <- methodsML("svmLinear", outcomeClass = "character")

set.seed(123)
trainedModel <- trainModel(
    inputData = scoresExample,
    metadata = pathMEDExampleMetadata,
    var2predict = "Response",
    models = modelsList,
    Koutter = 2,
    Kinner = 2,
    repeatsCV = 1
)

data(pathMEDExampleData, pathMEDExampleMetadata)

scoresExample <- getScores(pathMEDExampleData, geneSets = "tmod", 
                             method = "GSVA")

modelsList <- methodsML("svmLinear", outcomeClass = "character")

set.seed(123)
trainedModel <- trainModel(
    inputData = scoresExample,
    metadata = pathMEDExampleMetadata,
    var2predict = "Response",
    models = modelsList,
    Koutter = 2,
    Kinner = 2,
    repeatsCV = 1
)

Package 'pathMED'

Help Index

Annotate the pathways from a scores matrix

Description

Usage

Arguments

Value

Author(s)

References

See Also

Examples

Create a reference data object for input to the pathMED functions

Description

Usage

Arguments

Value

Author(s)

References

See Also

Examples

Split pathways into coexpressed subpathways

Description

Usage

Arguments

Value

Author(s)

References

See Also

Examples

Preloaded gene sets

Description

Usage

Format

Calculate pathways scores for a dataset

Description

Usage

Arguments

Value

Author(s)

References

See Also

Examples

Prepare the models parameter for the trainModel function

Description

Usage

Arguments

Value

Author(s)

References

See Also

Examples

Create a reference dataset based on M-scores

Description

Usage

Arguments

Value

Author(s)

References

See Also

Examples

Filter pathways from the reference M-scores dataset

Description

Usage

Arguments

Value

Author(s)

References

See Also

Examples

Estimate M-scores for a dataset without healthy controls

Description

Usage

Arguments

Value

Author(s)

References

See Also

Examples

Example of test gene expression data

Description