Prediction of MoRFs in Protein Sequences with MLPs Based on Sequence Properties and Evolution Information

Area of Science:

• Biochemistry
• Computational Biology
• Proteomics

Background:

• Intrinsically disordered proteins (IDPs) contain functional regions called molecular recognition features (MoRFs).
• MoRFs undergo disorder-to-order transitions upon binding to interaction partners.
• Accurate prediction of MoRFs is vital due to their association with various diseases and potential as drug targets.

Purpose of the Study:

• To develop and validate a novel computational method for predicting MoRFs.
• To leverage sequence properties and evolutionary information for enhanced MoRF prediction accuracy.
• To establish a competitive method for identifying disease-associated protein regions.

Main Methods:

• Development of two distinct multi-layer perceptron (MLP) neural networks.
• Implementation of a preprocessing strategy using sliding windows of varying sizes to capture diverse MoRF-related properties.
• Integration of Bayes rule with MLP outputs for final MoRF prediction.

Main Results:

• The developed method demonstrates competitive performance compared to existing state-of-the-art techniques.
• The combination of sequence properties and evolutionary information proves effective for MoRF prediction.
• The sliding window approach successfully captures relevant protein region characteristics.

Conclusions:

• The proposed computational method provides a reliable approach for predicting MoRFs.
• This method can aid in identifying potential drug targets for disease-related research.
• The findings contribute to the understanding and prediction of intrinsically disordered protein functions.