Hi Ning,
I am using the elbow method for clustering. Clustering is always subjective, so the simpler the better.
I am interested in what others have to say, especially the folks who made the program.
Thanks,
Valentin
Valentin V. Rybenkov
University of Oklahoma
101 Stephenson Parkway
Norman, OK 73019
Web: antibiotics.oucreate.com
From: ning via AMBER<mailto:amber.ambermd.org>
Sent: Friday, September 9, 2022 9:49 AM
To: amber maillist<mailto:amber.ambermd.org>
Subject: [AMBER] How to determine the optimal number of cluster in cpptraj or other something
Hi Amber experts,
I need to do a clustering analysis for a long time molecular dynamic simulation, about 2 microsecond. I had finished the cpptraj tutorial for clustering. The dbscan method may occupy too much memory storage and computation time. I choose the k-means method, whereas, I don't know how to determine the best number of clusters in trajectory processing. I read some papers that mentioned a method called "silhouette" which is able to calculate the differences between one cluster and one's neighboring cluster. I want to know how to acheive this function in CPPTRAJ tool. Any suggestions are welcome, please help me. Thanks in advance.
Sincerely,
Ning
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Sep 09 2022 - 08:00:03 PDT
