search

joandre / MCL_spark

An implementation of Markov Clustering algorithm for Spark in Scala
MIT License
34 stars 13 forks source link

Cluster converges in exactly one iteration for any data set #14

Open nishanthsiva opened 7 years ago

nishanthsiva commented 7 years ago

Hi,

I am using your project for gene clustering and I found that the clustering result is always the same irrespective of the data set I use and the parameters I provide as input.

Used this data set with varying inflation rate and iterations. The clustering results hardly change. Is there any documentation or a set of examples that define how the clustering works? I am not sure if this problem is because of my data set or my lack of understanding of the algorithm. examples.zip

joandre commented 7 years ago

Hi,

Thanks for using MCL package. Can you give me an example of inflation and expansion rate you used with a code example ? Did you use released version or did you build the package from up-to-date code?

evewurtele commented 7 years ago

Sorry for slowness of reply, my student Nishanth graduated just then. We used up to date code. We are going back through the parameters and will provide info. Can you share any further updates that are not yet on github? Are there any publications? We find your code really useful. Thanks much :)

joandre commented 7 years ago

No further updates right now, while I was waiting for an optimization of spark sparse matrix multiplication. But please take a look at akaltsikis repository : https://github.com/akaltsikis/Markov_Cluster_Algorithm.

joandre commented 7 years ago

And no problem for the late answer. :-) Congratulations to Nishanth. Glad to see it was useful.