Conformational Variability Prediction of Influenza Virus Hemagglutinins with Amino Acid Mutations Using...

Area of Science:

• Structural biology
• Bioinformatics
• Computational biology

Background:

• Supersecondary structure code (SSSC) uses conformation terms (H, S, T, D) to identify protein motifs from the Protein Data Bank (PDB).
• Deep neural networks offer advanced methods for predicting protein structure and function.

Purpose of the Study:

• To describe a protocol for conformational variability prediction using SSSC.
• To analyze influenza virus hemagglutinins and their mutations using this prediction method.

Main Methods:

• Utilizing a deep neural network-based system (SSSCPreds) for predicting protein flexibility/rigidity and regional shapes.
• Applying SSSC to analyze conformational variability patterns in influenza A (H1N1pdm and H5N1) hemagglutinins.
• Correlating predicted sequence flexibility/rigidity maps with sequence-to-phenotype data from mutations.

Main Results:

• SSSCPreds accurately predicts protein flexibility and rigidity, enabling correlation with mutation effects.
• Conformational variability patterns of H1N1pdm and H5N1 hemagglutinins show strong resemblance, differing mainly at the H5N1 furin cleavage site.
• Visualized conformational maps highlight virus variant transitions and increased flexibility, offering insights into influenza epidemics.

Conclusions:

• The SSSC-based conformational variability prediction method is accurate enough to correlate with sequence-to-phenotype data.
• This approach provides visual understanding of virus variant transitions and flexibility changes, relevant to influenza epidemics.
• A limitation exists for proteins at pH 5 due to limited measurement conditions affecting prediction accuracy.