Abstract :
[en] The knowledge of the tridimensional structure of a protein is essential to design drugs, to predict protein function and to study mechanism of protein function. 3D structure can be determined using two experimental techniques: X-Ray and NMR. However, these techniques have limitations: they are time consuming, manually intensive and sometime technically difficult. Due to these limitations, different approaches that combine the strength of computer and sparse experimental NMR data such as backbone chemical shifts, NOEs distances and residual dipolar couplings have been proposed for the determination of 3D protein structure. Among these experimental NMR data, backbone chemical shifts are the experimental NMR data that can be rapidly, easily and accurately measured. Thus different approaches that use sparse chemical shifts such as CS-Rosetta have been designed. Unfortunately, chemical shift-based 3D structure determination approaches are limited by the size and complexity of proteins (limited to molecular proteins that weight at most 15 kDa). This limitation can be understood by the fact that every folding trajectory during sampling is completely independent of every other. To overcome this limitation, additional sparse NMR data such as homology information, NOEs distances and/or RDCs are needed. CS-HM-Rosetta that combines incomplete chemical shifts and information derived from homologous structures have been developed in this purpose. CS-NOE-RDC-Rosetta, PHAISTOS have been also developed to guide structures determination using chemical shifts, NOEs distances and RDCs. Despite the fact that experimental data are invaluable for guiding sampling to the vicinity of the global energy minimum, for larger proteins, these data failed to guide sampling to the native minimum state. That is why, in a number of cases the improved sampling methodology makes a larger contribution than incorporation of additional experimental data. New sampling protocol named Resolution Adapted Structural RECombination (RASREC) has been therefore designed to overcome size limitation when using CS-Rosetta protocol. Unfortunately, most of these methods are not fully automated since they require manual assignment of experimental NMR data. Several automated methods such as AUDANA, CYANA 2015, AutoNOE-RASREC-CS-ROSETTA and J-UNIO that automatically assigned NOESY cross peaks, and chemical resonance assignment for J-UNIO have been developed. Up to now, they have not been compared together. In addition, they require expertise in computer science. Therefore, I am working toward getting fully automated approach, designed for non computer science and NMR experts that combines J-UNIO and AutoNOE-RASREC-CS-ROSETTA for rapid (4 weeks) protein 3D structure determination using unassigned backbone chemical shifts, NOESY spectra and RDCs spectra as input.
During the 15th edition of the YBMRS meeting, I will compare CS-Rosetta, CS-RASREC-Rosetta, CS-HM-Rosetta and Homology modeling approches based on the quality of derived 3D structures for proteins for which chemical shifts are available.
