Part 2B User Supplied Distance
Calculating the distance between Gene-Sets is by default calculating the RR. However in certain cases the user wants to use a different function to combine the data. The function allows to use a user supplied functions. e.g. the Jaccard similarity index can be used, this calculated the overlap between Gene-sets. Note: Jaccard is quite suitable for larger Gene-Sets but biased in smaller Gene-Sets.
The pipeline creates a matrix with Gene-Sets as rows and Genes as Columns. So any function supplied needs to accommodate the 2 different Gene Sets you want to calculate the distance in between needs to have an A and an B (or X and Y) that takes a vector with
GeneSet_1 = c(Gene_1 = 0,Gene_2 = 0,Gene_3 = 1,Gene_4 = 0,Gene_5 = 0,Gene_6 = 1,Gene_7 = 1,Gene_8 = 0)
GeneSet_2 = c(Gene_1 = 1,Gene_2 = 1,Gene_3 = 1,Gene_4 = 0,Gene_5 = 0,Gene_6 = 0,Gene_7 = 0,Gene_8 = 0)
Where in this case there 1 gene shared (Gene_3).
jaccard <- function(A,B)
{
#The Jaccard similarity index compares members
#for two sets to see which members are shared and which are distinct.
#It's a measure of similarity for the two sets of data, with a range from 0% to 100%.
#The higher the percentage, the more similar the two populations.
M <- sum(as.vector(A) == 1 & as.vector(B) == 1)
A.c <- sum(as.vector(A) == 1 & as.vector(B) == 0)
B.c <- sum(as.vector(A) == 0 & as.vector(B) == 1)
J <- M/(A.c+B.c+M)
return(J)
}
IPA.Object.J <- CombineGeneSets(Object = IPA.object1, combineMethod = "Jaccard", combineMethod.supplied = jaccard)
IPA.Object.J <- ClusterGeneSets(Object = IPA.Object.J,
clusters = 4,
method = "kmeans",
order = "group")
PlotGeneSets(Object = IPA.Object.J, fontsize =5,
legend = T,
annotation.mol=F,
main="Jaccard distance", RR.max = 50)
