Mean of Circular Quantities (MCQ)

Mean of Circular Quantities (MCQ) is a dissimilarity measure useful in comparison of 3D protein and/or RNA structures. It calculates an average difference between corresponding torsion angle values (rotations around bonds). More information can be found in:

Zok, T., Popenda, M., & Szachniuk, M. (2014). MCQ4Structures to compute similarity of molecule structures. Central European Journal of Operations Research, 22(3), 457–473. https://doi.org/10.1007/s10100-013-0296-5

Installation

git clone https://github.com/RNA-Puzzles/mcq-parent
cd mcq-parent
git submodule update --init
mvn install

Contents

This project consists of a few subprojets:

• mcq-common: base functionality
• mcq-clustering: partitional and hierarchical clustering
• mcq-cli: command-line interface
• mcq-gui: graphical interface

Main Ideas

• Use pl.poznan.put.comparison.MCQ#compareGlobally to compare two 3D structures and obtain a global value of dissimilarity. You can use pl.poznan.put.comparison.global.ParallelGlobalComparator to process multiple inputs in parallel
• Use pl.poznan.put.comparison.MCQ#comparePair to obtain detailed information about dissimilarity of two 3D structures
• Use pl.poznan.put.comparison.MCQ#compareModels in a situation where a distinguished reference 3D structure is known and you want to know how 3D models compare to it

Clustering

• Use pl.poznan.put.clustering.hierarchical.Clusterer to construct dendrograms from a distance matrix (with COMPLETE, SINGLE or AVERAGE linkage option)
• Use pl.poznan.put.clustering.partitional.KMedoids to perform partitional clustering based on distance matrix
• Use pl.poznan.put.clustering.partitional.KScanner#parallelScan to find optimum number of clusters with respect to silhouette score