immuneHallmarks.R

## Semantic similarity
library(GOSim)

l1 = list()
#l1$Adaptive_Response = c('GO:0002250','GO:0002819')
l1$Akt_Signalling = c('GO:0043491','GO:0051896')
l1$Antigen_Presentation = c('GO:0019882','GO:0002472','GO:0002625','GO:0019886')
l1$Apoptosis = c('GO:0006915','GO:0033028','GO:0071888','GO:2000110','GO:2000111','GO:0012501','GO:0048102') # ,'GO:2001242'
l1$Cell_Cycle = c('GO:0007050','GO:0051726','GO:0006260','GO:0000075')
l1$Cell_Cycle_Checkpoint = c('GO:0000075')
l1$M_Phase = c('GO:0000279')
l1$S_Phase = c('GO:0051320')
l1$G1_Phase = c('GO:0051318')
l1$G2_Phase = c('GO:0051319')
#l1$Interphase = c('GO:0051325')
#l1$Cell_Proliferation = c('GO:0008283') # GO:0061517
l1$Cytokine = c('GO:0050663','GO:0001816','GO:0010936','GO:0060907','GO:0032640')
l1$Innate_Response = c('GO:0006952','GO:0045087','GO:0002221','GO:0002775','GO:0046209','GO:0051001','GO:0051769','GO:0002237') # ,'GO:0035325','GO:0035663'
l1$Lymphocyte_Activation = c('GO:0046649')
l1$Mitochondrial = c('GO:0042775')
l1$Necrosis = c('GO:0070265') # ,'GO:0008220'
l1$Phagocytosis_Recognition = c('GO:0006910','GO:0006909','GO:0045728','GO:0010506')
#l1$TLR_Signalling = c('GO:0035661')

 
#d3 = getTermSim(l1$Mitochondrial,method='JiangConrath')
#d3[is.infinite(d3)] = NA
#d3[is.nan(d3)] = NA
#heatmap.2(as.matrix(d3),trace='none',col=colorpanel(256,'blue','black','yellow'),Colv=T,Rowv=T,density.info='none',margins=c(15,15),keysize=1) #,scale='row')

# To look at the overlap between terms
setOntology('BP')
# Testing overlap of terms we chose
hallmarks = matrix(ncol=length(names(l1)), nrow=length(names(l1)), dimnames=list(names(l1), names(l1)))
for(cluster in names(l1)) {
    for(hallmark in names(l1)) {
        if (!(cluster==hallmark)) {
            d2 = getTermSim(c(l1[[cluster]],l1[[hallmark]]),method='JiangConrath')
            d2[is.infinite(d2)] = NA
            hallmarks[cluster,hallmark] = max(d2[1:length(l1[[cluster]]),-(1:length(l1[[cluster]]))],na.rm=T)
        }
    }
}

heatmap.2(as.matrix(hallmarks),trace='none',col=colorpanel(256,'blue','black','yellow'),Colv=T,Rowv=T,dendrogram='none',density.info='none',margins=c(15,15),keysize=1) #,scale='row')
write.csv(hallmarks,'similarityOfHallmarks_VALL.csv')

# Doing the comparison with the enriched GOBP terms from meta-clusters
d1 = read.csv('clusterEnrichment_GOBP_classic_0.05_VALL.csv',header=T,row.names=1)
l2 = list()
for(cluster in rownames(d1)) {
    l2[[cluster]] = strsplit(as.character(d1[cluster,2]),';')[[1]]
}

hallmarks = matrix(ncol=length(names(l1)), nrow=length(names(l2)), dimnames=list(names(l2), names(l1)))
for(cluster in names(l2)) {
    if (!(length(l2[[cluster]])==0)) {
        for(hallmark in c('Cell_Cycle_Checkpoint')) {
        #for(hallmark in names(l1)) {
            d2 = getTermSim(c(l2[[cluster]],l1[[hallmark]]),method='JiangConrath')
            d2[is.infinite(d2)] = NA
            hallmarks[cluster,hallmark] = max(d2[1:length(l2[[cluster]]),-(1:length(l2[[cluster]]))],na.rm=T)
        }
    }
}

write.csv(hallmarks,'jiangConrath_hallmarks_classic_0.05_VALL.csv')