Clustering Tools

There are a few different sets of clustering tools. Some are in Packages/Clustering while others are python-based.

cluster-kgs

Performs hierarchical clustering, using an NMRClust-like method to determine the optimal number of clusters to retain. Takes a frame-to-frame distance matrix as input, eg the output of rmsds, multi-rmsds, or all_contacts.py. See also cluster_pops.py

cluster-structures.py

Performs k-means clustering on one or more trajectories, using RMSD as the metric. Has the option to do a t-SNE transformation first.

hierarchical-cluster.py

Performs hierarchical clustering given a distance matrix, using scipy's hierarchical clustering library. If you gave it multiple trajectories, the tool can report the cluster populations for each trajectory.

cluster_pops.py

Extracts the cluster populations from the output of cluster-kgs

frame-picker.py Create a trajectory for each cluster from cluster-kgs