FormatInput: Format the input for graph-based learning. This input...

In ncats/MultiOmicsGraphPrediction: Network-Based Ensemble Prediction from IntLIM Predictors

Format the input for graph-based learning. This input consists of: - The Laplacian of a line graph built from the co-regulation graphs, where each edge corresponds to a pair of analytes. - A prediction value for each edge of the line graph, for each sample X. - The true prediction values Y for each sample X. - The co-regulation graph. - The line graph built from the co-regulation graphs. - The input data (analyte levels, covariates, and phenotype). - The model results from IntLIM for each predictor. - The metafeatures for each model from IntLIM. - The phenotype / outcome column name in the input data. - The type of the outcome / phenotype ("numeric" or "categorical"). - The analyte type used as the outcome variable for each model (1 or 2). - The analyte type used as the independent variable for each model (1 or 2).

Description

Format the input for graph-based learning. This input consists of: - The Laplacian of a line graph built from the co-regulation graphs, where each edge corresponds to a pair of analytes. - A prediction value for each edge of the line graph, for each sample X. - The true prediction values Y for each sample X. - The co-regulation graph. - The line graph built from the co-regulation graphs. - The input data (analyte levels, covariates, and phenotype). - The model results from IntLIM for each predictor. - The metafeatures for each model from IntLIM. - The phenotype / outcome column name in the input data. - The type of the outcome / phenotype ("numeric" or "categorical"). - The analyte type used as the outcome variable for each model (1 or 2). - The analyte type used as the independent variable for each model (1 or 2).

Usage

FormatInput(
  predictionGraphs,
  coregulationGraph,
  metaFeatures,
  modelProperties,
  inputData,
  stype.class,
  edgeTypeList,
  stype,
  verbose = TRUE,
  covariates = c(),
  outcome = 2,
  independent.var.type = 2,
  errorCorrelationGroupReps
)

Arguments

predictionGraphs	A list of igraph objects, each of which includes predictions for each edge.
coregulationGraph	An igraph object containing the coregulation graph.
metaFeatures	The metafeature values computed during input prep (DoModelSetup()).
modelProperties	The results of RunIntLim().
inputData	An IntLimData object.
stype.class	The class of the outcome ("numeric" or "categorical")
edgeTypeList	List containing one or more of the following to include in the line graph: - "shared.outcome.analyte" - "shared.independent.analyte" - "analyte.chain"
stype	The phenotype or outcome of interest
verbose	Whether to print the number of predictions replaced in each sample. TRUE or FALSE. Default is FALSE.
covariates	A list of covariates to include in the model. These will be in the sampleMetaData slot of the inputData variable.
outcome	The outcome used in the IntLIM models.
independent.var.type	The independent variable type used in the IntLIM models.
errorCorrelationGroupReps	Representative predictors for groups of predictors with correlated error.

Value

An object of the ModelInput class. This object will contain the input data, graph information, and individual predictor properties and their metafeatures.

Slots

metaFeatures

A list of calculated meta-features for each sample

modelProperties

A data frame that includes model information, i.e. R^2, interaction term p-value, and coefficients.

inputData

An IntLIMData object that includes slots for the sample data, the analyte data, and the analyte meta data.

ncats/MultiOmicsGraphPrediction documentation built on Aug. 23, 2023, 9:19 a.m.