The low count cutoff in rank-based clustering plays a role in generating more
robust result for method such as k-means. However, when a continuous ordering
of enrichment is expected for method such as "total", this may generate
erroneous figures that are not very pretty. A smart choice should be used for
different clustering methods here.
Original issue reported on code.google.com by shenli.sam on 25 Aug 2014 at 2:40
Original issue reported on code.google.com by
shenli.samon 25 Aug 2014 at 2:40