EPNClassifier

EPNClassifier assigns ependymal tumors profiled by bulk transcriptomics (RNA-seq or microarrays) into molecular subgroups from Pajtler K.W. et al., Cancer Cell 27 (2015) and Pajtler K.W. et al., Acta Neuropathol 136 (2018). It is based on gene set enrichment analysis of pre-determined gene signatures. Significant p-values can be used to classify EPN tumors into their most-likely molecular subgroup. PFA tumors can be further classified into subtypes (PFA_1 and PFA_2). The training error of EPNClassifier is 1.4% and its test error across platforms is estimated to be <7%. More information abot EPNClassifier can be found in:

• Aubin R.*, Troisi E. C.*, Alghalith A. N., Nasrallah M. P., Santi M., and Camara P. G., "Cell Ecosystem and Signaling Pathways of Primary and Metastatic Pediatric Posterior Fossa Ependymoma". Submitted.

To cite this code:

Installation

devtools::install_github("CamaraLab/EPN_Classifier")
library(EPNClassifier)

Tutorial

Example data for 209 EPN tumors profiled by DNA microarrays in Pajtler K.W. et al., Cancer Cell 27 (2015) and can be downloaded here

Classify EPN into their molecular subgroups

Calculate the overall running sum statistics and p-vlaues for each EPN tumor.

classification <- ClassifyEPN(bulk = data, permutations = 10000)

#Assign each sample to a single molecular subgroup
molec_gps <- Classify(classification, min_pvalue = 0.35)

Plot the running sum statistic (ie. enrichment score ES) over a ranked list of genes for a specific EPN sample.

ES <- EnrichmentScore_EPN(bulk_sample = data[,1,drop=F])

par(mfrow=c(3,3))
for (i in 1:8){
  plot(x = ES[[i]]$x, y = ES[[i]]$y, xlab = "Rank List of Genes", ylab = "Running Sum Statistic", 
  main = paste0(names(ES)[i]," Enrichment Score"), cex=.1, ylim = c(-1,1) , panel.first = c(lines(x = ES[[i]]$x, 
  y = ES[[i]]$y, col = "red"),abline(h = 0)))
}

Classify PF_A tumors into their subtypes (PFA_1 and PFA_2)

Calculate the overall running sum statistics and p-vlaues for each PF_A tumor.

#Restrict to PFA tumors
pfa_data <- data[,names(molec_gps)[molec_gps %in% "PF_EPN_A"]]

classification_pf <- ClassifyPFA(bulk = pfa_data, permutations = 100000)

#Assign each sample to a single PFA subtype
pf_gps <- Classify(classification_pf, min_pvalue = 0.35)

Plot the running sum statistic (ie. enrichment score ES) over a ranked list of genes for a specific PF_A sample.

ES <- EnrichmentScore_PFA(bulk_sample = pfa_data[,1,drop=F])

par(mfrow=c(1,2))
plot(x = ES$PFA_1$x, y = ES$PFA_1$y, xlab = "Rank List of Genes", ylab = "Running Sum Statistic", 
  main = "PFA_1 Enrichment Score", cex=.1, ylim = c(-1,1) , panel.first = c(lines(x = ES$PFA_1$x, 
  y = ES$PFA_1$y, col = "red"),abline(h = 0)))
plot(x = ES$PFA_2$x, y = ES$PFA_2$y, xlab = "Rank List of Genes", ylab = "Running Sum Statistic", 
  main = "PFA_2 Enrichment Score", cex=.1, ylim = c(-1,1) , panel.first = c(lines(x = ES$PFA_2$x, 
  y = ES$PFA_2$y, col = "red"),abline(h = 0)))