Performing simulation time-segmentation using TICC
The following steps will generate a contacts for a simulation, use them as input to TICC clustered time-segmentation, and finally generate a clustermap and flareplot illustrating the differences between each cluster.
- The first step extracts atomic-level interactions.
- The second step computes TICC time-segment clusters and generates a residue-level frequency file for each cluster (here, k=3).
- The third step combines the frequency files into a fingerprint heatmap and a flareplot.
Assume you have a simulation represented with a topology file, top.pdb, and trajectory file,
trj.dcd. We first generate a contact list (contacts.tsv) and a residue contact frequency
for each cluster:
get_dynamic_contacts.py --topology top.pdb --trajectory trj.dcd --itypes hb --output contacts.tsv
get_contact_ticc.py --input_contacts contacts.tsv \
--clusters 3
--tab_output ticc.tsv
--frequency_output segments
The --frequency_output argument is used as a prefix for cluster-specific residue contact frequency
files:
- segments_resfreq_cluster000.tsv
- segments_resfreq_cluster001.tsv
- segments_resfreq_cluster002.tsv
Finally, the residue contact frequencies are combined into clustermap and flareplot comparison files:
get_contact_fingerprints.py --input_frequencies segments_resfreq_cluster*.tsv \
--column_headers Cluster0 Cluster1 Cluster2 \
--plot_output fingerprint.png \
--flare_output compareflare.json \