Recommended by

Indexed by

Atomic-level structure determination of amorphous molecular solids by NMR

Manuel Cordova1*, Pinelopi Moutzouri1, Sten O. Nilsson Lill2, Alexander Cousen3, Martin Kearns4, Stefan T. Norberg5, Anna Svensk Ankarberg5, James McCabe4, Arthur C. Pinon6, Staffan Schantz5*, Lyndon Emsley1*

1 Institut des Sciences et Ingénierie Chimiques, École Polytechnique Fédérale de Lausanne (EPFL), CH-1015 Lausanne, Switzerland

2 Data Science & Modelling, Pharmaceutical Sciences, R&D, AstraZeneca, Gothenburg, Sweden

3 Early Chemical Development, Pharmaceutical Sciences, R&D, AstraZeneca, Macclesfield, UK

4 Early Product Development and Manufacturing, Pharmaceutical Sciences, R&D, AstraZeneca, Macclesfield, UK

5 Oral Product Development, Pharmaceutical Technology & Development, Operations, AstraZeneca, Gothenburg, Sweden

6 Swedish NMR Center, Department of Chemistry and Molecular Biology, University of Gothenburg, 41390 Gothenburg, Sweden

* Corresponding authors emails: manuel.cordova@epfl.ch, Staffan.Schantz@astrazeneca.com, lyndon.emsley@epfl.ch
DOI10.24435/materialscloud:gk-51 [version v1]

Publication date: Jul 25, 2023

How to cite this record

Manuel Cordova, Pinelopi Moutzouri, Sten O. Nilsson Lill, Alexander Cousen, Martin Kearns, Stefan T. Norberg, Anna Svensk Ankarberg, James McCabe, Arthur C. Pinon, Staffan Schantz, Lyndon Emsley, Atomic-level structure determination of amorphous molecular solids by NMR, Materials Cloud Archive 2023.112 (2023), doi: 10.24435/materialscloud:gk-51.


Structure determination of amorphous materials remains challenging, owing to the disorder inherent to these materials. Nuclear magnetic resonance (NMR) powder crystallography is a powerful method to determine the structure of molecular solids, but disorder leads to both a high degree of overlap between measured signals, resulting in challenges for spectral assignment, and prevents the unambiguous identification of a single modelled periodic structure as representative of the whole material. Here, we determine the atomic-level ensemble structure of the amorphous form of the drug AZD4625 by combining solid-state NMR experiments with molecular dynamics (MD) simulations and machine-learned chemical shifts. By considering the combined shifts of all 1H and 13C atomic sites in the molecule, we determine the structure of the amorphous form by identifying an ensemble of local molecular environments that are in agreement with experiment. We then extract preferred conformations and intermolecular interactions in the amorphous sample, and analyze the structure in terms of the hydrogen bonding and conformational factors that stabilize the amorphous form of the drug.

Materials Cloud sections using this data

No Explore or Discover sections associated with this archive record.


File name Size Description
352.7 MiB Raw NMR data of the experimental spectra measured.
1009.0 MiB PDB files containing the snapshots extracted from the MD simulations of the amorphous sample.
807.2 MiB Numpy arrays containing the chemical shifts predicted using ShiftML2 on the MD snapshots.
147.5 MiB JSON files containing the geometric features extracted from the MD snapshots.
24.3 MiB JSON and Numpy files containing the scores obtained for molecular environments from the MD snapshots to match the NMR experiments.
286.3 MiB XYZ files of 8000 randomly selected molecular environments from the MD snapshots. Each environment is represented by three files, "_cen" for the central molecule, "_env" for the neighbouring molecules, and "_all" for the complete environment.
359.8 MiB XYZ files of the 10107 NMR-selected molecular environments from the MD snapshots. Each environment is represented by three files, "_cen" for the central molecule, "_env" for the neighbouring molecules, and "_all" for the complete environment.
302.3 MiB Input and output files of the DFTB energy computation performed on the randomly selected molecular environments.
382.2 MiB Input and output files of the DFTB energy computation performed on the NMR-selected molecular environments.
104.1 KiB Raw data of the pairwise distribution function (PDF) measurements of the amorphous sample in Gudrun format.
197.9 KiB Numpy arrays containing the simulated pairwise distribution function (PDF) from the MD snapshots and NMR-selected molecular environments.
14.1 MiB Interaction maps obtained for amorphous AZD4625
726.1 KiB XYZ and VMD files used to visualize the structure of amorphous AZD4625.
154.7 KiB Python libraries necessary to run the analysis scripts.
336.4 KiB Jupyter notebooks containing the Python scripts used to analyse the data.


Files and data are licensed under the terms of the following license: Creative Commons Attribution Share Alike 4.0 International.
Metadata, except for email addresses, are licensed under the Creative Commons Attribution Share-Alike 4.0 International license.

External references

Journal reference (Paper in which the method is described)
M. Cordova, P. Moutzouri, S. O. Nilsson Lill, A. Cousen, M. Kearns, S. T. Norberg, A. Svensk Ankarberg, J. McCabe, A. C. Pinon, S. Schantz, L. Emsley, in press (2023)


MARVEL/DD1 NMR Amorphous solids machine learning molecular dynamics simulation structure determination

Version history:

2023.112 (version v1) [This version] Jul 25, 2023 DOI10.24435/materialscloud:gk-51