8 days ago by
EMBL Heidelberg, Germany
The way to deal with large numbers of anything is to have a computer do the work. The question you should ask yourself is what are you trying to assess ? There are many ways to evaluate a clustering result but they depend on the context and the goal of the clustering. If you have anything resembling a ground truth, then use it. You can also use feature enrichment analysis (e.g. any annotation you may have) to assess whether the clusters make sense in the context of the experiments. Otherwise you can compute various indices such as the silhouette.