ShiftML: chemical shifts in molecular solids by machine learning

About ShiftML

ShiftML was developed jointly by the Ceriotti and Emsley groups at the EPFL. ShiftML uses a machine learning framework to predict chemical shifts in solids which is based on capturing the local environments of individual atoms. ShiftML has been trained on 2000 structures taken from the Cambridge Structural Database (CSD), chosen to be as diverse as possible. On a separate test set of 500 randomly selected structures it predicts chemical shifts of molecular solids to an accuracy with respect to GIPAW DFT of (RMSE) 0.49 ppm for 1H, 4.3 ppm for 13C, 13.3 ppm for 15N, and 17.7 ppm for 17O and with an R2 of 0.97 for 1H, 0.99 for 13C, 0.99 for 15N, and 0.99 for 17O.

Please read and cite the following: Paruzzo, F. M., Hofstetter, A., Musil, F., De, S., Ceriotti, M., & Emsley, L. (2018). Chemical shifts in molecular solids by machine learning. Nature Communications, 9(1), 4501.

Version 1.02 Beta; 27.06.2018

This is the first release of ShiftML, with limited functionality. This version predicts chemical shifts of structures containing only 1H, 13C, 15N and 17O. It has only been tested for input structures in which all atomic coordinates are fully relaxed with DFT. Version 1.02 Beta introduces sparsified kernels, which reduces the calculation time and memory requirement while maintining similar accuracy.

What ShiftML does

ShiftML takes as an input a crystal structure (in a number of different formats), and

  • predicts the isotropic NMR chemical shieldings/shifts of 1H, 13C, 15N and 17O;
  • provides a JSmol interactive output, and copy-paste content in the extended xyz format, in the magres format (the calculated shieldings/shifts are given as a tensor with the isotropic shielding/shift as all three diagonal elements) and in JSON.

ShiftML has so far been tested for DFT-optimized crystal structures of molecular solids containing 1H, 13C, 15N and 17O atoms. Paruzzo, et al..

Version 1.02 Beta implements a sparsified multi-scale-kernel version, which reduces the calculation time and memory requirement while maintining similar accuracy.

Version history

17.09.2018: Version 1.03 Beta

  • Change the prediction procedure to reduce the memory requirements of the app.

27.06.2018: Version 1.02 Beta

  • Implementation of a sparsified multi-scale-kernel, which reduces the calculation time and memory requirement while maintining similar accuracy.

02.05.2018: Version 1.0 Beta

  • Implementation of the multi-scale-kernel version based on the description in Paruzzo, et al..

Upload your structure

  Chemical shieldings
  Chemical shifts





By continuing, you agree with the terms of use of this service.

Please be patient.
Depending on the workload of the server the calculation might only start in a few minutes.

Otherwise, pick an example

Please be patient.
Depending on the workload of the server the calculation might only start in a few minutes.

How to cite

If you use this tool, please cite the following work:

Acknowledgment

We are gratefull for the support from: