Session ThOB. There are 4 abstracts in this session.

Session: SMALL MOLECULES 2, time: 08:30 - 8:55 am

Free Ligand 1D NMR Conformational Signatures to aid medicinal chemistry design of potent drug candidates

Elisabetta Chiarparin1; Amber Balazs2
1AstraZeneca, Cambridge, UK; 2AstraZeneca, Waltham, MA

A range of drug discovery case studies demonstrate the power of free ligand NMR conformational analysis to enhance Structure Based Drug Design. Here we will show how to extract unique ligand conformational signatures from 1D NMR spectra, routinely utilised by the chemists for the determination of compound identity and purity. Case studies from recent AstraZeneca oncology research projects, will be used to demonstrate how 1D NMR conformational signatures synergise with measured potency data to rationalise structure activity relationships (SAR), and with physicochemical end-points to rationalise properties driving oral absorption. Learning to extract free ligand conformational information from routinely available 1D NMR spectra has proven to be fast enough to effectively guide medicinal chemistry decisions within design cycles for compound optimization.

Session: SMALL MOLECULES 2, time: 08:55 - 9:10 am

Rapid crystal structure determination via machine learned chemical shifts

Martins Balodis1; Albert Hofstetter1; Manuel Cordova1; Graeme Day2; Lyndon Emsley1
1EPFL, Lausanne, Switzerland; 2University of Southampton, Southampton, United Kingdom

Recently an ultra-fast chemical shift prediction method, ShiftML, has emerged that allows to predict isotropic chemical shifts in less than a second using machine learned chemical shifts for molecular solids. It opens the possibility to introduce accurate, on the fly chemical shift calculations in the structure determination protocols.

As a proof of concept, here we successfully determine the crystal structure of cocaine through prior unbiased constraints from solid-state NMR, and using on the fly generated isotropic chemical shifts to guide structure determination starting from a randomly generated gas phase conformers.

Session: SMALL MOLECULES 2, time: 09:25 - 9:40 am

Overcoming Resolution Limits in Crowded Spectra by Automated Global Fitting to Analytical NMR Lineshapes

Joshua Dudley; Sojeong Park; Emanual Fetene; Colin Smith
Wesleyan University, Middletown, CT

A longstanding problem facing small-molecule and biomolecular NMR is overlapped signals in crowded spectra. Nearly all methods for separating overlapped signals model peaks using Lorentzian, Gaussian, or mixtures of both functions. However, due to signal truncation, apodization, and unresolved scalar couplings, such peak shapes miss important features, such as truncation artifacts, present in real NMR signals. To address this, we have derived analytical models that precisely account for these effects. Together with new automated algorithms that can globally fit multiple spectra, this enables extraction of parameters such as R2* and scalar couplings that were previously inaccessible. We show that it is possible to accurately determine unresolved protein backbone scalar couplings. The method is implemented in the open-source FitNMR R package.

Session: SMALL MOLECULES 2, time: 09:40 - 9:55 am

Diffusion selective pulses

Zhaoyuan Gong; Jamie D. Walls
Department of Chemistry, University of Miami, Coral Gables, FL 33146

In NMR and MRI, differences in self-diffusion coefficients are routinely exploited to exponentially suppress the signals from fast diffusing species relative to the signals from slow diffusing species. However, such diffusion-based spectral editing is unable to only suppress signals from species with a particular diffusion coefficient. In this work, diffusion selective pulses are presented that can selectively suppress signals for only a particular value of self-diffusion coefficient. Experimental demonstrations of diffusion selective pulses are provided in water and water/acetone/dimethylsulfoxide samples.