Gapped sequence alignment using artificial neural networks: application to the MHC class I system.

Area of Science:

• Immunology
• Computational Biology
• Bioinformatics

Background:

• Biological processes often involve receptor interactions with variable-length linear ligands, such as Major Histocompatibility Complex (MHC) class I molecules binding peptides.
• MHC class I alleles typically exhibit length preferences for bound peptides, generally ranging from 8 to 11 amino acids.

Purpose of the Study:

• To develop an advanced sequence alignment method utilizing artificial neural networks capable of handling insertions and deletions.
• To enhance the prediction accuracy of peptide-MHC class I binding affinity by incorporating variable peptide lengths.

Main Methods:

• Development of a novel sequence alignment method based on artificial neural networks that accommodates insertions and deletions.
• Training and evaluation of prediction models using alignments that include variable peptide lengths.

Main Results:

• Prediction methods incorporating insertions and deletions demonstrated significantly higher performance compared to those trained on fixed-length peptides.
• The analysis of deletion locations provided insights into peptide-MHC binding mechanisms, including peptide bulging and terminal protrusions.
• The developed method successfully learned the length-specific binding profiles of different MHC molecules, reducing the experimental effort for epitope discovery.

Conclusions:

• The gapped alignment method significantly improves peptide-MHC binding prediction accuracy.
• This approach offers valuable insights into peptide-MHC binding modes and facilitates more efficient epitope identification.