predForm-Site: Formylation site prediction by incorporating multiple features and resolving data imbalance

Area of Science:

• Biochemistry
• Bioinformatics
• Computational Biology

Background:

• Formylation is an emerging post-translational modification of lysine residues.
• Dysregulated formylation is implicated in various disease pathologies.
• Accurate identification of formylation sites is crucial for understanding its biological roles.

Purpose of the Study:

• To develop a novel, highly accurate computational tool for predicting formylation sites.
• To improve the representation of formylation sites using integrated sequence features.
• To optimize the prediction model for enhanced performance on imbalanced datasets.

Main Methods:

• Development of a novel predictor named predForm-Site.
• Integration of multiple sequence features for informative site representation.
• Optimization of classifier decision functions on skewed datasets during training.

Main Results:

• predForm-Site achieved 99.5% sensitivity, 99.8% specificity, and 99.8% accuracy (AUC 0.999) in jackknife tests.
• Independent testing yielded >97% sensitivity and >99% specificity.
• Outperformed CKSAAP_FormSite in sensitivity (~20%), specificity (~30%), and accuracy (~22%).

Conclusions:

• predForm-Site demonstrates superior performance in identifying formylation sites.
• The tool serves as a valuable complementary resource for formylation site exploration.
• An accessible online version is available for the scientific community.