A topological similarity measure for proteins

• We introduce a novel method for assessing similarity among chemical structures.
• The method is built on well-established computational topology algorithms.
• First, a barcode representation of the topology of molecules is calculated.
• We determine similarity by comparing these barcodes in a rigorous manner.
• We validate our method by comparing molecules from different sources.

We introduce a new measure for assessing similarity among chemical structures, based on well-established computational-topology algorithms. We argue that although the method considers geometry, it is more than a mere geometric similarity measure, as it takes into account, on different geometric scales, the important topological features of the compared structures. We prove that our measure is rigorous and complies with the proper mathematical requirements. We validate the method through comparing different configurations of simple zinc finger proteins and present an application on ligands binding to membrane-proteINS extracted from the Directory of Useful Decoys: Enhanced database and corresponding decoys. This article is part of a Special Issue entitled: Viral membrane proteins — Channels for cellular networking.