regression_selected_pathways: regression_selected_pathways

In TaoDFang/GENEMABR: Gene-set Enrichment with Regularized Regression

Description

This function allows you to extracte enriched pathways for gene module/list via regressioin (elastic net) based method

Usage

regression_selected_pathways(gene_input, gene_pathway_matrix = NULL,
  alpha = 0.5, family = c("gaussian", "binomial"), lambda = NULL,
  verbose = TRUE, ...)

Arguments

gene_input	A vecor of genes to be annotated. It should have same ID types(Ensembl ID, HUGO gene symbol) as the genes in gene_pathway_matrix.
gene_pathway_matrix	A binary background matrix whose columns are the pathways/gene sets and whose rows are all the genes from pathways/gene sets . It could be in sparse matrix format ((inherit from class "sparseMatrix" as in package Matrix) to save memory. For gene i and pathway j, the value of matrix(i,j) is 1 is gene i belonging to pathway j otherwise 0. Users could leave it as default value then it will use pre-collected gene_pathway_matrix from GO Ontology and REACTOME databaase. Otherwise, they could use their own customized gene_pathway_matrix
alpha	The elasticnet mixing parameter, with 0~≤q~α~≤q~1. The penalty is defined as (1-α)/2||β||_2^2+α||β||_1. alpha=1 is the lasso penalty, and alpha=0 the ridge penalty. Default value: 0.5.
family	Response type, currently gaussian and binomial are supported and the gaussian family is the default. Future extensions are likely.
lambda	A user supplied lambda sequence, see glmnet and use with care.
verbose	If supprese warning messesge from gerr functions. TRUE or FALSE. Default value TRUE, not suppressing warning messages.
...	Other paramaters passed to the cv.glmnet function.

Value

A list of following elements:

• selected_pathways_names - Pathways names for selected pathways

• selected_pathways_coef - Regression coefficients value for selected pathways

• selected_pathways_fisher_pvalue - Fisher exact pvalue for selected pathways

• selected_pathways_num_genes - The number of genes for selected pathways in background

• model - the trained glmnet model

• x - independent variable

• y - dependent variable

Examples

rspResults <- regression_selected_pathways(gene_input=c("TRPC4AP","CDC37",
  "TNIP1","IKBKB","NKIRAS2", "NFKBIA","TIMM50","RELB","TNFAIP3","NFKBIB",
  "HSPA1A","NFKBIE","SPAG9","NFKB2","ERLIN1","REL","TNIP2",
  "TUBB6","MAP3K8"),
 gene_pathway_matrix=NULL,lambda=NULL,alpha=0.5)

TaoDFang/GENEMABR documentation built on July 28, 2019, 3:16 p.m.