CTLPred Algorithm

  • General Information about T cell Epitopes
  • The new paradigm in vaccine design is emerging, following essential discoveries in immunology and development of new T-cell epitope prediction tools.T cells orchestrate the immune response against the intracellular and extracellular pathogens.T cells regulate the immune response by accelerlating and deaccelerating the antibody production. These cells themselves act as a killer for the pathogens.The T cells are not able to recognise the antigen in the native form.They recognise the processed antigen which is presented on surface of Antigen Presenting Cells (APCs).There are different pathways for processing and presentation of extracellular and interacellular pathogens.

  • Briefly about Antigen processing & presentation
  • The T cell epitopes are derived from the intracellular pathogens by firstly delivering to protein degrading machinery known as proteasome.This leads to the fragmentation of the antigenic protein.Some of these peptides are transported to endoplasmic reticulum through a special transporter known as TAP protein (transporter assiociated antigen presentation).In the last step, the peptides are loaded to the MHC molecules for presntation on the surface of APC. It is observed that only 0.05% of peptides transported to ER bind to MHC class I molecules.The MHC class I molecules have a groove cloased at both ends for peptide, so they mostly binds to 8-10 amino acid long peptides.The crucial steps for T cell epitopes or more specifically CTL epitopes identification is the production of peptide fragments by protease,transport of peptides to ER and binding of the peptides to MHC molecules. The rules for these events have been described making it possible to formulate the prediction algorithms.The data about all of these crucial steps is not present in such amount that computer can make generalised rules.Very little amount of data is available for TAP and proteasomal peptides.The more data of these crucial peptides is prerequisite because the accuracy of the prediction algorithms is solely the reflection of quality and quantity of data.

    The extracellular pathogens are endocytosed by the various cells. These pathogens are fragmented in the special compartment of these cells which are acidic and rich in the proteases and acidic enviorment.The peptides generated in this compartment compete for binding to MHC molecules and HLA-DM acts to facilitate this process.MHC class II molcules transport these peptides to surface. The MHC class II binds to 11-25 amino acids long peptides.

  • Extraction and preprocessing of data of CTL epitopes and non-epitopes.
  • Development of ANN based prediction method.
  • Development of SVM based prediction method.
  • Consensus prediction based on SVM and ANN based methods
  • Combined prediction based on SVM and ANN based Prediction

SNNSSVMConsensus
EpitopeEpitopeEpitope
EpitopeNon-epitopeNon-epitope
Non-epitopeEpitopeNon-epitope
Non-epitopeNon-epitopeNon-epitope

    Combined Prediction based on SVM and ANN prediction:-

    In combined prediction the peptides which are precdicted epitopes by either of the prediction methods(SVM or ANN) or by both methods, are considered as predicted epitopes otherwise they are considered as non-epitopes. This method will result in improvment of senstivity as well as NPV of prediction.

SNNSSVMConsensus
EpitopeEpitopeEpitope
EpitopeNon-epitopeepitope
Non-epitopeEpitopeepitope
Non-epitopeNon-epitopeNon-epitope

This demonstrates that ANN and SVM extract different patterns from the data.So, the consensus or combined prediction by simultaneous use of the both prediction methods is more accurate.