HiCglmiModel: The generalized linear model with interactions.

In HiCglmi: Probing Factor-Dependent Long-Range Contacts using Regression with Higher-Order Interaction Terms

Description

The main function to compute the generalized linear model with interactions.

Usage

HiCglmiModel(hrpd, model, facModel, regressionMode = "NB", includeBias = T, 
		orient = F, sampleSize = NULL, distInter = NULL)

Arguments

hrpd	Preprocessed data. It includes Hi-C interaction pairs that have been annotated with genomic feature information. To compute this object, you should use the function HiCglmiProcData.
model	A formula object (stats R package). See Section HiCglmi-package, for examples.
facModel	A vector of character describing the genomic features. For instance, facModel="BEAF32".
regressionMode	If regressionMode=="Poisson", a Poisson regression. If regressionMode=="QP", a quasi Poisson regression. If regressionMode=="NB", a negative binomial regression. If regressionMode=="Poisson", a Poisson lasso regression.
includeBias	If includeBias==True, then GC, mappability and fragment length biases are included in the regression (default mode). If includeBias==False, biases are not included in the regression. You can used this option if you are using Hi-C matrix that had been previously corrected for biases (not recommended).
orient	If orient=True, then interaction variable is computed using orientation on DNA. This is used for DNA motif analysis. For instance, to show long-range contacts between CTCF motifs in convergent orientation.
sampleSize	Optional. A numerical value for subsampling Hi-C data.
distInter	Optional. A vector of two numerical values to set a distance range for Hi-C analysis.

Value

If regressionMode is "Poisson", "QP" or "NB", then a summary(glm) object is returned.

Author(s)

Raphael Mourad

HiCglmi documentation built on Aug. 20, 2019, 1:04 a.m.