Closed Tima-Ze closed 1 year ago
Hi Tima, right, thank you for noticing this. We changed the algorithm sometime ago and didn't reflect the changes in the vignette. I fixed it in a75afc9c6a79a07abb7dbfb2623230acd019588b and 35e2117429beb83af522ad50ca796a3bd463521e, hopefully it is clear now.
The right meaning of the threshold parameter should be "reduceSimMatrix
groups terms which are at least within a similarity below threshold
".
You can fetch the changes from github (devtools::install_github("ssayols/rrvgo")
) or from the Bioconductor devel branch which I guess will be available in a couple of days.
Hi Sergi, Regarding
reduceSimMatrix()
functionAccording to the manual document the threshold actually determines/represents the similarity. However, it seems the opposite to be the case. In the source code :
cluster <- cutree(hclust(as.dist(1 - simMatrix), method), h = threshold)
, where the threshold is based on dissimilarity matrix, so lower thresholds indicate a greater similarity between clusters. Am I right, or did I just get misunderstood? Best, Tima