Computerized Analysis of 2D NMR Correlation Spectra Using Discrete Matrix Representations

Abstract

Methodology is described for generation and analysis of experimental correlation spectra as discrete matrix representations (DMR). DMRs are constructed by equating ωi and ωj for each correlation to individual rows and columns in discrete matrices. Experimental connectivity patterns are explicitly encoded by assigning integer labels, equal to the peak labels of experimental correlations, to the individual matrix elements. These labels may be keyed to additional information (e.g., intensity, linewidths, or personal comments) pertaining to individual correlations. Such data reduction is demonstrated with simulated and experimental data obtained for a 27-residue peptide. The DMR of an experimental HOHAHA spectrum may be block diagonalized to reveal blocks of correlations belonging to individual spin systems. Such blocks provide the basis for recognition and resolution of correlations belonging to individual spin systems. These blocks are subsequently interrogated to obtain frequency lists associated with different spin systems. These lists are in turn used to analyze complementary correlation spectra (e.g., COSY or NOESY) for the same sample. DMRs provide a convenient format for computerized representation, manipulation, and analysis of 2D NMR correlation data. They also permit forms of data abstraction and pattern recognition unavailable with traditional connectivity path searches.