Identifying Human SIRT1 Substrates by Integrating Heterogeneous Information from Various Sources.

Area of Science:

• Biochemistry
• Computational Biology
• Molecular Biology

Background:

• Protein acetylation is a vital post-translational modification impacting biological activities.
• Sirtuin 1 (SIRT1), a Class III HDAC, plays a key role in acetylation, but its function is difficult to elucidate due to a limited number of known substrates.

Purpose of the Study:

• To develop a novel computational method for screening and identifying SIRT1 substrates.
• To enhance the accuracy of SIRT1 substrate prediction by incorporating diverse protein features.

Main Methods:

• A Support Vector Machine (SVM) based computational approach was designed.
• The method integrates both primary protein sequence and functional features for prediction.
• A manually curated dataset was used for training and validation.

Main Results:

• Integrating functional features improved the Matthews correlation coefficient (MCC) from 0.10 to 0.65.
• The developed classifier demonstrated high effectiveness in identifying SIRT1 substrates.
• Validation using an independent dataset and biological experiments confirmed the method's accuracy.

Conclusions:

• The novel computational method effectively identifies SIRT1 substrates, aiding in understanding SIRT1's biological roles.
• The tool provides a valuable resource for filtering candidate substrates for further experimental research.
• An online prediction tool is available to support the scientific community.