The Treedist distance matrix output of Generalized RF and Nye et al. methods are zero

[Jigyasa3]

Dear @ms609,

Thank you again for the detailed manual and explanation of the methods! I do have a question on ClusteringInfoDistance() function and NyeSimilarity() functions.

I am running the following functions-

for distance matrix-

tree1<-read.tree(file="hosttree-d__Bacteria_p__Desulfobacterota_COG0215_tips_1.nwk")
tree2<-read.tree(file="symbionttree-d__Bacteria_p__Desulfobacterota_COG0215_tips_1.nwk")
tree1<-unroot(tree1)

#GRF
dist_rf <- ClusteringInfoDistance(tree1, tree2, normalize = TRUE)

#Nye
dist_ny <- NyeSimilarity(tree1, tree2, normalize = TRUE ,similarity = FALSE)

for p-values-

#GRF
nRep <- 100000 # Use more replicates for more accurate estimate of expected value
randomTrees <- lapply(logical(nRep), function (x) RandomTree(tree1$tip.label))
randomDists <- ClusteringInfoDistance(tree1, randomTrees, normalize = TRUE)
expectedCID <- mean(randomDists)

dist12 <- ClusteringInfoDistance(tree1, tree2, normalize = TRUE)
# Now count the number of random trees that are this similar to tree1
nThisSimilar <- sum(randomDists < dist12)
pValue <- nThisSimilar / nRep

#Nye-
nRep <- 100000 # Use more replicates for more accurate estimate of expected value
randomTrees <- lapply(logical(nRep), function (x) RandomTree(tree1$tip.label))
randomDists <- NyeSimilarity(tree1, randomTrees, normalize = TRUE,similarity = FALSE)
expectedCID <- mean(randomDists)

dist12 <- NyeSimilarity(tree1, tree2, normalize = TRUE,similarity = FALSE)
# Now count the number of random trees that are this similar to tree1
nThisSimilar <- sum(randomDists < dist12)
pValue2 <- nThisSimilar / nRep

I am getting a zero distance matrix and p-value outputs for the trees attached. Tree1-https://github.com/Jigyasa3/errors/blob/master/hosttree-d__Bacteria_p__Desulfobacterota_COG0215_tips_1.nwk and Tree2- https://github.com/Jigyasa3/errors/blob/master/symbionttree-d__Bacteria_p__Desulfobacterota_COG0215_tips_1.nwk. The two trees are completely identical to each other, yet the value of the distance matrix is 0. Why do you think that's happening?

Looking forward to your reply!

[ms609]

If the trees are identical (notwithstanding edge lengths, which are ignored), then they should have a distance of zero, and a high similarity. Distance increases as trees become less similar.

Or have I misunderstood your question?

[Jigyasa3]

Dear @ms609 ,

Thanks for a quick reply! May I please confirm again that if I use the function ClusteringInfoDistance() with normalize = TRUE then lower the value of the distance means higher similarity?

Similarly, if I use the function NyeSimilarity() with normalize = TRUE , similarity = FALSE then lower the value of the distance also means higher similarity?

The normalize = TRUE would help to compare the two distances with each other so that I can say that if the ClusteringInfoDistance() gives me zero while NyeSimilarity(similarity=FALSE) gives me 0.2 then the two trees are very similar to each other via both the methods.

#GRF
dist_rf <- ClusteringInfoDistance(tree1, tree2, normalize = TRUE)

#Nye
dist_ny <- NyeSimilarity(tree1, tree2, normalize = TRUE ,similarity = FALSE)

[ms609]

Yes, that's right: you need the similarity = FALSE argument to ask NyeSimilarity() to return a distance (i.e. difference); then the interpretation of the two distance measures is equivalent (though the absolute values will differ).