nmrgnn-data 0.7

Chemical shift prediction dataset

Data for NMR GNN

This contains the parsing scripts and data used for our GNN chemical shift predictor model.

Install

pip install nmrgnn-data

Numpy Error

If you see this error:

ValueError: numpy.ndarray size changed, may indicate binary incompatibility. Expected 88 from C header, got 80 from PyObject

Try re-install numpy

pip uninstall -y numpy && pip install numpy

Parsing Scripts

To install with the parsing functionality, use this

conda install -c omnia -c conda-forge rdkit openmm numpy==1.18.5
pip install nmrgnn-data[parse]

Working with Data

All commands below can have additional information printed using the --help argument.

Find pairs

Find pairs of atoms with chemical shifts that are neighbors and sort them based on distance.

nmrdata find-pairs structure-test.tfrecords-data.tfrecord ALA-H ALA-N

Count Names

Get class/atom name counts:

nmrdata count-names structure-test.tfrecords-data.tfrecord

Validate

Check that records are consistent with embeddings

nmrdata validate-embeddings structure-test.tfrecords-data.tfrecord

Check that neighbor lists are consistent with embeddings

nmrdata validate-nlist structure-test.tfrecords-data.tfrecord

Check that peaks are reasonable (no nans, no extreme values, no bad masks)

nmrdata validate-peaks structure-test.tfrecords-data.tfrecord

Output Lables

To extract labels ordered by PDB and residue:

nmrdata write-peak-labels test-structure-shift-data.tfrecord  test-structure-shift-record-info.txt labels.txt

Making New Data

See commands parse-shiftml, parse-metabolites, parse-shiftx which are parsers for various databases.

From RefDB Files

This requires a pickled python object called data.pb to be in the directory. It is a list of dicts containing pdb_file (path to PDB), pdb (PDB ID), corr (path to .corr file), and chain (which chain). chain can be _ to indicate use first chain.

nmrparse parse-refdb directory name --pdb_filter exclude_ids.txt

Citation

Please cite Predicting Chemical Shifts with Graph Neural Networks

@article{yang2021predicting,
  title={Predicting Chemical Shifts with Graph Neural Networks},
  author={Yang, Ziyue and Chakraborty, Maghesree and White, Andrew D},
  journal={Chemical Science},
  year={2021},
  publisher={Royal Society of Chemistry}
}