CompDTUReg: Implement the CompDTU and CompDTUme regression model

In skvanburen/CompDTUReg: CompDTUReg: Fit Compositional Regression Models for DTU

Implement the CompDTU and CompDTUme regression model

Description

CompDTUReg runs the CompDTU and CompDTUme regression models for a specific gene.

Usage

CompDTUReg(
  genename,
  Y = NULL,
  Group = NULL,
  runWithME = TRUE,
  YInfRep = NULL,
  mean.withinhat = NULL,
  customHypTest = FALSE,
  NullDesign = NULL,
  AltDesign = NULL
)

Arguments

genename	The name of the current gene to run the method on.
Y	corresponds is the ilr transformed matrix of response values for the non-inferential replicate data. Matrix should have a number of rows corresponding to the number of samples and a number of columns corresponding to the number of ilr coordinates (which one less than the number transcripts remaining after filtering). Rows need to be in the same order as either Group or the design matrices specified by AltDesign and NullDesign. Set to NULL if you are running the model with measurement error, as it is not used.
Group	A vector of condition assignments corresponding to the samples. Needs to be ordered the same as Y or YInfRep. Set to NULL if custom design matrices will be specified by the NullDesign and AltDesign arguments.
runWithME	is TRUE/FALSE indicating whether the model should be run with measurement error or not (corresponding to CompDTU and CompDTUme models respectively). If runWithME is TRUE ensure YInfRep is non-NULL and if runWithME is FALSE ensure Y is non-NULL.
YInfRep	corresponds is the ilr transformed matrix of response values for the inferential replicate data. Matrix should have a number of rows corresponding to the number of samples times the number of replicates and a number of columns corresponding to the number of ilr coordinates (which is one less than the number transcripts remaining after filtering). Rows must be ordered as Sample1InfRep1, Sample2InfRep1, Sample3Infrep1, etc. NOT Sample1InfRep1, Sample1InfRep2, Sample1InfRepe, etc. Set to NULL if you are running the model without measurement error, as it is not used.
mean.withinhat	is the mean of the sample-specific covariance matrices of the inferential replicates (calculated on the ilr scale). The condition variable should not be included in this matrix because it is included automatically via the Group parameter. Number of rows needs to be the number of samples, even if CompDTUme is to be run (the matrix will be replicated automatically as needed). Set to NULL if custom design matrices will be specified by the NullDesign and AltDesign arguments.
customHypTest	should be set to TRUE if custom design matrices will be specified using the NullDesign and AltDesign arguments. If TRUE the Group argument must be set to NULL. Default is FALSE.
NullDesign	specifies the design matrix corresponding to the null hypothesis if a custom hypothesis test is specified via customHypTest being set to TRUE. Number of rows needs to be the number of samples, even if CompDTUme is to be run (the matrix will be replicated automatically as needed).
AltDesign	specifies the design matrix corresponding to the alternative hypothesis if a custom hypothesis test is specified via customHypTest being set to TRUE. Number of rows needs to be the number of samples, even if CompDTUme is to be run (the matrix will be replicated automatically as needed).

Details

This function is run separately for each gene and the easiest way to run it will be to follow the pipeline given in the SampleCode folder of the package. In particular if gene-level files created from SaveGeneLevelFiles are used the easiest way to use this is to use the helper function startCompDTUReg, which loads the gene-level results. These gene-level files contain all input arguments aside from any custom specified null and alternative hypothesis design matrices, and input arguments are automatically set by startCompDTUReg. See the file (4)RunCompositionalRegressions.R in the package's SampleCode folder for example code.

Value

a data.frame containing the gene_id being used, p-value from the CompDTU or CompDTUme significance test for condition, and various information on the current dataset being used.

skvanburen/CompDTUReg documentation built on Jan. 23, 2025, 9:01 a.m.