Predicting Amyloid Formation in Antibodies Using AI
Author Information
Author(s): David Maria Pamela C, Concepcion Gisela P, Padlan Eduardo A
Primary Institution: University of the Philippines Diliman
Hypothesis
Can simple artificial intelligence methods accurately predict the amyloidogenicity of antibody sequences?
Conclusion
The study suggests that both naive Bayesian classifiers and decision trees can effectively predict the amyloidogenicity of antibody sequences.
Supporting Evidence
- The naive Bayesian method achieved an accuracy of 60.84% for predicting amyloidogenic sequences.
- Augmenting the training set improved the accuracy to 81.08%.
- The decision tree method yielded an average prediction accuracy of 78.64%.
Takeaway
This study used computer programs to help figure out if certain antibodies might form harmful clumps in the body. It found that using more examples helps make better predictions.
Methodology
The study employed naive Bayesian classifiers and decision trees to analyze 143 amyloidogenic and 158 non-amyloidogenic antibody sequences.
Potential Biases
Potential overfitting due to the limited sample size and the complexity of the models used.
Limitations
The study's predictions are limited by the small number of available sequences for training the classifiers.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website