Related Experiment Video
Updated: Jul 7, 2026
06:19
High-Throughput Screening to Obtain Crystal Hits for Protein Crystallography
Published on: March 10, 2023
ParCrys: a Parzen window density estimation approach to protein crystallization propensity prediction
Ian M Overton1, Gianandrea Padovani, Mark A Girolami
1School of Life Sciences Research, University of Dundee, Dow Street, Dundee, DD1 5EH and Department of Computing Science, University of Glasgow, Glasgow, GL12 8QQ, UK.
Bioinformatics (Oxford, England)
|February 21, 2008
Summary
We developed ParCrys, a new computational method to predict protein crystallization success. ParCrys (Parzen Window approach) aids structural biology and genomics by ranking proteins likely to form diffraction-quality crystals.
Area of Science:
- Structural Biology
- Computational Biology
- Structural Genomics
Background:
- Predicting protein crystallization is crucial for structural biology and high-throughput structural genomics.
- Accurate prediction of crystallization success aids in prioritizing protein targets.
Purpose of the Study:
- To introduce ParCrys, a novel computational method for estimating protein crystallization propensity.
- To evaluate ParCrys performance against existing methods using established databases.
Main Methods:
- Utilized a Parzen Window approach for predicting protein crystallization.
- Employed data from the Protein Data Bank (PDB) for training.
- Used TargetDB and PepcDB for feature selection and independent testing.
Main Results:
- ParCrys demonstrated superior performance compared to OB-Score, SECRET, and CRYSTALP.
- Achieved an accuracy of 79.1% and a Matthews correlation coefficient of 0.582.
- Maintained strong performance (74.0% accuracy, 0.227 MCC) on 'real-world' imbalanced datasets.
Conclusions:
- ParCrys offers a more accurate prediction of protein crystallization success.
- The method is a valuable tool for structural genomics and structural biology initiatives.
- Predictions and data are publicly available for further research.
