EZ-ASSIGN, a program for exhaustive NMR chemical shift assignments of large proteins from complete or incomplete triple-resonance data
- 418 Downloads
For several of the proteins in the BioMagResBank larger than 200 residues, 60 % or fewer of the backbone resonances were assigned. But how reliable are those assignments? In contrast to complete assignments, where it is possible to check whether every triple-resonance Generalized Spin System (GSS) is assigned once and only once, with incomplete data one should compare all possible assignments and pick the best one. But that is not feasible: For example, for 200 residues and an incomplete set of 100 GSS, there are 1.6 × 10260 possible assignments. In “EZ-ASSIGN”, the protein sequence is divided in smaller unique fragments. Combined with intelligent search approaches, an exhaustive comparison of all possible assignments is now feasible using a laptop computer. The program was tested with experimental data of a 388-residue domain of the Hsp70 chaperone protein DnaK and for a 351-residue domain of a type III secretion ATPase. EZ-ASSIGN reproduced the hand assignments. It did slightly better than the computer program PINE (Bahrami et al. in PLoS Comput Biol 5(3):e1000307, 2009) and significantly outperformed SAGA (Crippen et al. in J Biomol NMR 46:281–298, 2010), AUTOASSIGN (Zimmerman et al. in J Mol Biol 269:592–610, 1997), and IBIS (Hyberts and Wagner in J Biomol NMR 26:335–344, 2003). Next, EZ-ASSIGN was used to investigate how well NMR data of decreasing completeness can be assigned. We found that the program could confidently assign fragments in very incomplete data. Here, EZ-ASSIGN dramatically outperformed all the other assignment programs tested.
KeywordsLarge protein NMR Computer assignment Assignment verification
E.R.P.Z. acknowledges support from National Institutes of Health grant NS059690 (to G.E. Gestwicki, P.I.). I.B. was supported by National Institutes of Health grant HL 102662 (to S. Ragdale, P.I.); P.R. was supported by National Institutes of Health grant AI094623 (to C. G. Kalodimos, P.I.). The authors acknowledge N.K. Khanra (Rutgers) for the type III ATPAse sample preparation and helpful discussions.
- Goddard TD, Kneller DG (2000) SPARKY 3. University of California, San FranciscoGoogle Scholar
- Zimmerman DE, Kulikowski CA, Huang Y, Feng W, Tashiro M, Shimotakahara S, Chien C, Powers R, Montelione GT (1997) Automated analysis of protein NMR assignments using methods from artificial intelligence. J Mol Biol 269:592–610Google Scholar