Speaker
Description
Clustering conformations of molecules or molecular assemblies is a
central problem faced in most studies, be they concerned with atomic
models or coarse grain models. This talk will review recent
contributions in this realm.
The first one is concerned with the newly designed molecular distance
RMSDcomb [1]. Given the decomposition of a structure into subdomains,
RMSDcomb provides a weighted average of the lRMSD observed between
these subdomains, stressing the role of local similarities--as opposed
to the lRMSD which suffers from the lack of homogeneity inherent to
global structural comparisons.
The second one is concerned with clustering methods providing a simple
read out for the number of clusters [2]. In this perspective, we will
briefly review density based clustering and the associated
stability assessment based on topological persistence.
Finally, we shall discuss the first technique finding a many-to-many
correspondence between two sets of clusters delivered by two different
methods or the same method under two sets of parameters [3], which is
particularly useful to consolidate competing clusterings.
All the methods discussed are provided within the Structural
Bioinformatics Library, in the following packages:
https://sbl.inria.fr/doc/Molecular_distances_flexible-user-manual.html
https://sbl.inria.fr/doc/Cluster_engines-user-manual.html#sec-Cluster_engines-programs
https://sbl.inria.fr/doc/D_family_matching-user-manual.html
[1] F. Cazals, T. Dreyfus, D. Mazauric, A. Roth and C.H. Robert.
Conformational Ensembles and Sampled Energy Landscapes: Analysis and
Comparison, J. Comp. Chem., 36, 2015.
[2] F. Cazals and R. Tetley. Characterizing molecular flexibility by
combining lRMSD measures, Proteins, 87, 2019.
[3] F. Cazals, D. Mazauric, R. Tetley and R. Watrigant.
Comparing two clusterings using matchings between clusters of clusters,
ACM J. of Experimental Algorithms, 24, 2019.
