RaptorX-Property: a web server for protein structure property prediction

Area of Science:

• Computational biology
• Bioinformatics
• Structural bioinformatics

Background:

• Predicting protein structure properties from sequence is crucial for understanding protein function.
• Existing methods often struggle with proteins lacking homologous structures in the Protein Data Bank (PDB) or with limited evolutionary information.
• Accurate prediction of secondary structure, solvent accessibility, and disorder regions remains a challenge.

Purpose of the Study:

• To introduce RaptorX Property, a new web server for predicting protein structure properties.
• To evaluate the performance of the RaptorX Property server, particularly for proteins with sparse sequence profiles.
• To demonstrate the effectiveness of the DeepCNF deep learning model in predicting protein sequence-structure relationships.

Main Methods:

• Development of the RaptorX Property web server.
• Implementation of the Deep Convolutional Neural Fields (DeepCNF) deep learning model.
• Prediction of secondary structure (SS), solvent accessibility (ACC), and disorder regions (DISO) from protein sequences without using templates.
• Evaluation on benchmark datasets including CASP10 and CASP11.

Main Results:

• RaptorX Property achieved high prediction accuracies: ~84% Q3 for 3-state SS, ~72% Q8 for 8-state SS, and ~66% Q3 for 3-state ACC.
• The server demonstrated strong performance in predicting disorder regions with an area under the ROC curve (AUC) of ~0.89.
• The DeepCNF model effectively captures complex sequence-structure relationships and interdependencies between adjacent property labels.

Conclusions:

• RaptorX Property is a powerful and accurate tool for predicting protein structure properties, especially for sequences with limited evolutionary information.
• The DeepCNF model represents a significant advancement in deep learning for protein structure prediction.
• The server outperforms existing methods, offering valuable insights into protein structure and function.