RF-GlutarySite: a random forest based predictor for glutarylation sites.

Area of Science:

• Biochemistry
• Computational Biology
• Proteomics

Background:

• Glutarylation is a newly identified posttranslational modification on lysine residues.
• It plays a role in metabolic and mitochondrial processes.
• Identifying specific glutarylation sites and proteome-wide extent is challenging.

Purpose of the Study:

• To develop a computational method for predicting glutarylation sites from primary amino acid sequences.
• To complement expensive and time-consuming proteomic analyses.
• To facilitate the study of glutarylation's characteristics and functional impacts.

Main Methods:

• Utilized Random Forest (RF) machine learning strategies.
• Identified key physiochemical and sequence-based features correlated with glutarylation.
• Developed and validated the RF-GlutarySite algorithm using 10-fold cross-validation and an independent test set.

Main Results:

• RF-GlutarySite achieved high accuracy (75%), sensitivity (81%), specificity (68%), and MCC (0.50) in 10-fold cross-validation.
• Independent test set validation showed comparable scores: 72% ACC, 73% SN, 70% SP, and 0.43 MCC.
• The tool demonstrated superior sensitivity compared to existing glutarylation site predictors.

Conclusions:

• RF-GlutarySite is an effective and efficient tool for predicting glutarylation sites.
• The method can uncover novel glutarylation sites and aid in understanding its relationships with other lysine modifications.
• This predictor facilitates research into the functional roles of glutarylation in biological processes.