How to evaluate different clustering methods for single cell RNA-seq

Question

[kaji331]

Hi, everyone:

I found that there was no golden standard method for single cell RNA-seq subgroup. There were many papers showed different methods for this analysis. Now I wanna try these different clustering methods to my data. How to evaluate them using R, python, etc. I don't want to do biological experiments~

Generally, is there a golden standard (methods, books, etc.) for evaluating clustering methods by computer in machine learning area?

Thanks very much !

Answer 1

[Jean-Karim Heriche]

There is no standard method for evaluating clustering methods other than comparing to some ground truth when it is available. There are however several measures of clustering "quality" e.g. the silhouette (in R silhouette). See for example the python sklearn page on evaluating clustering.