Support faster clustering algorithms


For our group the bottleneck at the moment for an analysis is the clustering part, especially for Illumina data. Are there plans to support other clustering methods which are faster like Esprit-Tree or the recently published Clustom, besides the cluster.split option? See the article here:



Sure, we’d love to incorporate new algorithms… assuming they’re good and based on a sequence alignment. ESPRIT is not good and incorporation of average neighbor makes the program explode. The Clustom paper makes no mention of assessing a confusion matrix to indicate the quality of the clustering. These would be the criteria that a new method would have to satisfy.