treeclimbR
treeclimbR is an algorithm to pinpoint the data-dependent resosultion on hierarchcial hypotheses.
Installation
devtools::install_github("fionarhuang/treeclimbR")Basic example (here)
We first generate a random tree with 100 leaves, each leaf representing an entity. Then we sample counts of entities in each sample from a multinomial distribution. In total, there are 40 samples (20 in group A and the other 20 in group B). 18 entities are randomly selected to have their counts multiplied by 2 in group A.
To run treeclimbR, we perform wilcoxon sum rank test on all nodes of the tree to obain P-values and directions. Details are here. Candidates proposed at different t values are shown in the animation below.
Heatmap shows counts of entities (rows) in samples (columns) split by groups. Branches that include entities with signals are colored in orange.
treeclimbR vs BH
Nodes identified by treeclimbR are compared to those identified by BH under FDR 0.05.
More scenarios (here)
- BS
- SS
Learn more
More examples about using treeclimbR could be found here