TCGAvisualize_SurvivalCoxNET: Survival analysis with univariate Cox regression package (dnet)
Description
TCGAvisualize_SurvivalCoxNET can help an user to identify a group of survival genes that are
significant from univariate Kaplan Meier Analysis and also for Cox Regression.
It shows in the end a network build with community of genes with similar range of pvalues from
Cox regression (same color) and that interaction among those genes is already validated in
literatures using the STRING database (version 9.1).
TCGAvisualize_SurvivalCoxNET perform survival analysis with univariate Cox regression
and package (dnet) using following functions wrapping from these packages:
is a data.frame using function 'clinic' with information
related to barcode / samples such as bcr_patient_barcode, days_to_death ,
days_to_last_followup , vital_status, etc
dataGE
is a matrix of Gene expression (genes in rows, samples in cols) from TCGAprepare
Genelist
is a list of gene symbols where perform survival KM.
org.Hs.string
an igraph object that contains a functional protein association network
in human. The network is extracted from the STRING database (version 10).
scoreConfidence
restrict to those edges with high confidence (eg. score>=700)
titlePlot
is the title to show in the final plot.
Value
net IGRAPH with related Cox survival genes in community (same pval and color) and with
interactions from STRING database.
query <- TCGAquery(tumor = "lgg")
Details
TCGAvisualize_SurvivalCoxNET allow user to perform the complete workflow using coxph
and dnet package related to survival analysis with an identification of gene-active networks from
high-throughput omics data using gene expression and clinical data.
Cox regression survival analysis to obtain hazard ratio (HR) and pvaules
fit a Cox proportional hazards model and ANOVA (Chisq test)
Network comunites
An igraph object that contains a functional protein association network in human.
The network is extracted from the STRING database (version 9.1).
Only those associations with medium confidence (score>=400) are retained.
restrict to those edges with high confidence (score>=700)
extract network that only contains genes in pvals
Identification of gene-active network
visualisation of the gene-active network itself
the layout of the network visualisation (fixed in different visuals)
color nodes according to communities (identified via a spin-glass model and simulated annealing)