Benchmarking pKa Prediction Methods
Author Information
Author(s): Matthew N Davies, Christopher P Toseland, David S Moss, Darren R Flower
Primary Institution: Edward Jenner Institute for Vaccine Research
Hypothesis
Can different software programs accurately predict pKa values for proteins?
Conclusion
PROPKA is the most accurate method for predicting pKa values, especially for surface residues.
Supporting Evidence
- PROPKA produced the fewest outliers and was computationally faster than other programs.
- MCCE was better for buried residues, while PROPKA excelled for surface residues.
- The study used a larger dataset than previous studies to evaluate the predictive capabilities of the programs.
Takeaway
This study looked at different computer programs to see which one is best at guessing how acidic or basic parts of proteins are. PROPKA did the best job!
Methodology
The study analyzed a dataset of experimentally-determined pKa values using various prediction techniques and statistical analyses.
Potential Biases
The performance of the programs varied significantly based on the residue type and its spatial location.
Limitations
Technical difficulties with the UHBD program limited the analysis to only 43 proteins from the dataset.
Participant Demographics
The dataset included 100 proteins with known pKa values.
Statistical Information
P-Value
0.05
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website