SiteBinder – software tool for superimposing multiple structural motives

R. Svobodová Vařeková^1,2, D. Sehnal¹, L. Pravda¹, J. Oppelt¹, and J. Koča¹

¹National Centre for Biomolecular Research, Faculty of Science, Masaryk University, Kamenice 5

625 00 Brno-Bohunice, Czech Republic

²ANF DATA, a Siemens company, Pujmanové 1221/4, 140 00, Praha 4, Czech Republic

svobodova@chemi.muni.cz

A lot of information about 3D structure of proteins (including metalloproteins, lectines etc.) is nowadays available in the Protein Data Bank database. This large amount of data allows us to analyze biochemically significant motives in these molecules. For example, we can study binding sites of metals in metalloproteins, surroundings of sugars in lectines, or other biochemically important motives (such approaches were used for example in [1] and [2]).

To perform these analyses a software tool is needed, which is capable of superimposing many structural motives at once (minimizing their mutual root mean square deviation - RMSD) and computing a model of the particular motive. The structural motives often exhibit partial symmetry; therefore, even for two motives, there exist several valid pairings of atoms of the structures, each yielding a good value of RMSD. Nevertheless, when superimposing multiple structures at once, the chosen pairing plays a vital role and different pairings might result in different values of RMSD. Unfortunately, currently available tools for superimposing of molecules (VMD, UCSF Chimera, etc.) are mainly focused on whole molecules and assume the pairing of atoms is given a priori. There are more sophisticated tools such as PyMol that optimize pairing; nevertheless, they automatically neglect poorly fitting atoms, which is an undesirable feature when trying to compute a model of the motive.

Therefore, a software package called SiteBinder was designed and implemented. This software applies currently published algorithms for alignment using quaternion math [3] and for superimposing of multiple structures [4] (both algorithms were adapted for our needs). Furthermore, SiteBinder provides algorithms for searching for optimal pairing of atoms (which utilizes the information about residues, atomic names, and chemical symbols contained in PDB file format) and calculating a model from multiple structures. SiteBinder allows the user to select subset of atoms on each structure which are then used for superimposing. This is achieved through user friendly and intuitive interface with features such as real-time highlighting of atoms of interest and automatic selection of atoms on multiple structures at once. Furthermore, detailed information about RMSD is provided including average RMSD to the other structures in the ensemble (this feature is very useful for example for classification of subclasses of a given structural motives). Finally, SiteBinder is able to export the superimposed structures in PDB format as well as in a rendered image.

We tested the software on several sets of biochemically important motives (i.e. about 500 Zinc finger motives, about 1000 Ca binding sites in lectines etc.) and compared RMSD values with other software (VMD, UCSF Chimera, and PyMol). SiteBinder provided better results for most of the tested motives.

1. F.P. Casey, N.E. Davey, I. Baran, R. Svobodová Vařeková, D.C. Shields: Web Server To Identify Similarity of Amino Acid Motifs to Compounds (SAAMCO), J. Chem Inf Model, 10 (2008) 1524 - 1529.

2. I. Baran, R. Svobodová Vařeková, L. Parthasarathi, S. Suchomel, F. Casey, D.C. Shields: Identification of Potential Small Molecule Peptidomimetics Similar to Motifs in Proteins. J Chem Inf Model, 47 (2007) 464 - 474

3. E.A. Coutsias, C. Seok, K.A. Dill: Using Quaternions to Calculate RMSD. J Comp Chem, 25 (2004) 1849 - 1857

4. X. Wang, J. Snoeyink: Defining and Computing Optimum RMSD for Gapped and Weighted Multiple-Structure Alignment. IEEE/ACM Trans Comput Biology Bioinform, 5 (2008), 525 – 533