The efficacy and safety of protein therapeutics depends critically on their High Order Structure (HOS), and changes to HOS during manufacture or storage can render them inactive or promote dangerous immune responses. Methods to measure and characterize HOS are essential for development of new biotherapeutics; for evaluating less expensive “biosimilar” replacements; and for monitoring and improving manufacturing, formulation, and stability. Nuclear magnetic resonance (NMR), which can provide detailed information on structure and dynamics at atomic resolution, is a powerful tool to probe HOS, but typical biomolecular applications use isotopic enrichment, long measurement times, and require extensive and often subjective interactive analysis by an expert.
Our computational methods development has two primary goals. The first goal is continued support of expert-driven biomolecular structure determination by NMR, with an emphasis on spectral reconstruction and quantification. The second goal is to develop computational alternatives to interactive analysis and assignment of spectral features, to provide practical HOS characterization of protein therapeutics via chemometrics and machine learning that is both objective and automated.
Ying J, Delaglio F, Torchia DA, Bax A: Sparse multidimensional iterative lineshape-enhanced (SMILE) reconstruction of both non-uniformly sampled and conventional NMR data. Journal of Biomolecular NMR. doi:10.1007/s10858-016-0072-7, 2016
Long D, Delaglio F, Sekharm A, Kay LE: Probing Invisible, Excited Protein States by Non-Uniformly Sampled Pseudo-4D CEST Spectroscopy. Angewandte Chemie International Edition in English 54: 10507-10511, 2015
NMR; Structural biology; Spectral processing; Spectral analysis; Spectral fingerprinting; Biotherapeutics; Higher order structure; Computation; Software;