NMR spectral analysis using prior knowledge

International Meeting on High-Dimensional Data-Driven Science 2015

Kasai, Takuma^1,2 ; Nagata, Kenji^3 ; Okada, Masato^3 ; Kigawa, Takanori^1,2,4

Laboratory for Biomolecular Structure and Dynamics, Cell Dynamics Research Core, RIKEN Quantitative Biology Center, Yokohama, Japan^1

JST CREST, Yokohama, Japan^2

Department of Complexity Science and Engineering, Graduate School of Frontier Sciences, University of Tokyo, Kashiwa, Japan^3

Department of Computational Intelligence and Systems Science, Interdisciplinary Graduate School of Science and Engineering, Tokyo Institute of Technology, Yokohama, Japan^4

关键词: Amino acid sequence;    Incorporating prior knowledge;    Nuclear Magnetic Resonance (NMR);    Protein structures;    Selective labeling;    Sequential assignment;    Spectral deconvolution;    Stable-isotope labeling;   

Others  :  https://iopscience.iop.org/article/10.1088/1742-6596/699/1/012003/pdf
DOI  :  10.1088/1742-6596/699/1/012003

来源: IOP

【 摘 要 】

Signal assignment is a fundamental step for analyses of protein structure and dynamics with nuclear magnetic resonance (NMR). Main-chain signal assignment is achieved with a sequential assignment method and/or an amino-acid selective stable isotope labeling (AASIL) method. Combinatorial selective labeling (CSL) methods, as well as our labeling strategy, stable isotope encoding (SiCode), were developed to reduce the required number of labeled samples, since one of the drawbacks of AASIL is that many samples are needed. Signal overlapping in NMR spectra interferes with amino-acid determination by CSL and SiCode. Since spectral deconvolution by peak fitting with a gradient method cannot resolve closely overlapped signals, we developed a new method to perform both peak fitting and amino acid determination simultaneously, with a replica exchange Monte Carlo method, incorporating prior knowledge of stable-isotope labeling ratios and the amino-acid sequence of the protein.