Publication details

Accurate Fitting SAXS Curves with NMR Structure Ensembles

Authors

KŘENEK Aleš KUBÍČEK Karel ŠTEFL Richard FILIPOVIČ Jiří

Year of publication 2015
Type Article in Proceedings
Conference Proceedings of International Symposium on Grids and Clouds 2015
MU Faculty or unit

Institute of Computer Science

Citation
Web http://pos.sissa.it/archive/conferences/239/025/ISGC2015_025.pdf
Field Biochemistry
Keywords saxs; nmr; ensamble fit
Description Typical NMR analyses of a biomolecule yields a set of up to few dozens candidate 3D structures of the analyzed molecule without any clues to discriminate among them further. A parallel SAXS experiment on the same sample can be used for this purpose. Previous implementations of “ensemble fit” (search for a mix of molecular conformations which matches the SAXS curve) were designed to choose from a huge ensemble generated by molecular dynamics. Therefore the methods must trade off accuracy for manageable speed, and they end up in mixing curves computed with rather different values of parameters which have physical meaning, which should be avoided. On the contrary, with a relatively small input set of candidate NMR structures we take a more accurate approach. Both the model parameters, considered globally now, and weights of individ- ual candidate structures (reflecting their presence in the solution) become independent variables of a multidimensional global optimization problem; the optimized value is the accuracy of the fit to the experimental data. The optimization must escape from traps of many local minima there- fore we use Monte Carlo with stochastic tunnelling. The method also offers opportunities for parallelization. The final issue is user friendliness of the entire workflow, which is quite complex, involving several programs to be run, handling different file formats, and setting multiple parameters, ending up with visualization of results. We outline design of a web portal hiding these complexities to the end user.
Related projects:

You are running an old browser version. We recommend updating your browser to its latest version.

More info