NVR-BIP: Nuclear Vector Replacement using Binary Integer Programming for NMR Structure-Based Assignments

Abstract

Nuclear magnetic resonance (NMR) spectroscopy is an important experimental technique that allows one to study protein structure and dynamics in solution. An important bottleneck in NMR protein structure determination is the assignment of NMR peaks to the corresponding nuclei. Structure-based assignment (SBA) aims to solve this problem with the help of a template protein which is homologous to the target and has applications in the study of structure–activity relationship, protein–protein and protein–ligand interactions. We formulate SBA as a linear assignment problem with additional nuclear overhauser effect constraints, which can be solved within nuclear vector replacement’s (NVR) framework (Langmead, C., Yan, A., Lilien, R., Wang, L. and Donald, B. (2003) A Polynomial-Time Nuclear Vector Replacement Algorithm for Automated NMR Resonance Assignments. Proc. the 7th Annual Int. Conf. Research in Computational Molecular Biology (RECOMB), Berlin, Germany, April 10–13, pp. 176–187. ACM Press, New York, NY. J. Comp. Bio., (2004), 11, pp. 277–298; Langmead, C. and Donald, B. (2004) An expectation/maximization nuclear vector replacement algorithm for automated NMR resonance assignments. J. Biomol. NMR, 29, 111–138). Our approach uses NVR’s scoring function and data types and also gives the option of using CH and NH residual dipolar coupling (RDCs), instead of NH RDCs which NVR requires. We test our technique on NVR’s data set as well as on four new proteins. Our results are comparable to NVR’s assignment accuracy on NVR’s test set, but higher on novel proteins. Our approach allows partial assignments. It is also complete and can return the optimum as well as near-optimum assignments. Furthermore, it allows us to analyze the information content of each data type and is easily extendable to accept new forms of input data, such as additional RDCs.

Incorporating Amino Acid Typing Into Nuclear Magnetic Resonance Protein Structure-Based Assignments

“…NVR-BIP uses a BIP solver to find the minimum scoring assignment. NVR-BIP was tested on 7 proteins with 25 templates and resulted in higher accuracies than NVR-EM [4,6].…”

“…Then various methods (such as Monte Carlo Simulation, memetic algorithm or integer programming) are employed to find the assignments corresponding to the optimum or near-optimum of this scoring function (see e.g. MONTE [2], MATCH [3], NVR-BIP [4]). …”

“…In [4], the authors developed a tool called NVR-BIP which can be used to solve the SBA problem. NVR-BIP uses the Nuclear Vector Replacement (NVR) framework [5,6] with additional sources of data, to determine the scoring function, and binary integer programming (BIP) to find the assignment.…”

“…In NVR-BIP, the assignment problem is formulated as an integer linear model with additional Nuclear Overhauser Effect (NOE) constraints. The authors presented their results on several proteins [4].…”

Assigning NMR spectra of irregular RNAs by heuristic algorithms

“…Several tools supporting the interpretation of NMR spectral data for proteins and algorithms dedicated to their assignment have been published. They employ different algorithmic approaches, such as best-first search [5], exhaustive and heuristic search [6][7][8] branch-and-bound [9], simulated annealing [10], genetic algorithms [11], Monte Carlo optimization [12,13], and other [14,15]. Despite the fact that a lot has been already done in the area, and -what is more -further computation provides for protein structures showing the quality comparable with this of models solved using X-ray crystallography [16], the assignment procedure is still a weak point of structure elucidation process.…”