MMBPred Algorithm


The server predicts the mutated promiscuous and high affinity MHC binding peptides.The server utilizes the matrix data in a linear prediction model, in which peptide conformation is entirely neglected. The only parameters describing peptide binding are peptide side chain effects on binding and peptide length are used for prediction.

The viral proteins are often lacking antigenic peptides.It was observed that peptides of these viral proteins are not able to form stable MHC peptides complex.Sette et al., 1994 observed that immunogenicity and MHC binding affinity peptides have direct relationship.This means the increase in affinity of MHC binding peptides may result in enhancement of immunogenicity of peptides.This enhancement in affinity of peptides toward MHC can be achieved by introducing mutations in natural peptides. The achievment of a optimal mutation experimentally requires a large number of experiments which are very cumbersome and time consuming. The computional prediction of such mutations provides the best alternative.This server allows the prediction of mutated high affinity binders having mutation at one, two or three positions. Another obstacle in subunit vaccine is that it is not effective in all population due to MHC polymorphism.The MHC polymorphism means that different MHC molecules binds to slightly different peptides.TO overcome the barrier of MHC restriction there is requirment of such peptides which can bind to many MHC alleles. These peptides are known as promiscuous MHC binders. This server also predicts promiscuous mutated MHC binding peptides crucial for a subunit vaccine designing.

The prediction is based on the quantitative matrices. The server can run prediction for 47 MHC class I alleles.
Generation of Quantitative matrices:

The quantitative matrices for 47 MHC alleles have been generated in this study. The MHC binder data of 9 amino acids for each MHC alleles were obtained from MHCBN database (A comprehensive database of MHC binders and non-binders). The data of non-binders for each MHC allele is also obtained from MHCBN database (wherever available) otherwise the nonamer peptides are randomly chosen from SWISS-PROT database. The dataset of each MHC allele have equal number of MHC binders and non-binders.

The cofficient values of all amino acids at positions (1 to 9) of blocks of 9 amino acids was calculated as follows:
i)The probability (Pb) of each amino acid from position 1 to 9 in binder dataset is calculated by applying standard probability calculating method on complete dataset.
ii)The probability (Pnb) of each amino acid from position 1 to 9 in non-binder dataset is calculated by following standard probability calculating method on complete dataset.
iii)The coefficient value is obtained by subtracting the Pnb from Pb.
In the generation of quantitative matrices the anchor residues were given more weightage. The anchor residues for specific MHC allele has been obtained from the work of rammensee et al., 1995. The quantitative matrices are generated for 47 alleles by following the above-described steps.

Threshold Score: The calculation of threshold or cutoff score is crucial for matrix based prediction The stringency of predictions can be controlled by varying the threshold or cutoff score. Thus, it is important to calculate the threshold score for each allele so that binders and non-binders can be discriminated. Ideally, one needs sufficient number of binders and non-binders to calculate the threshold score. The lack of peptides particularly non-binders make its impossible to calculate the threshold score. In order to overcome this problem we have adopted strategy followed in ProPred (Singh and Raghava , 2001).for each matrix.

i)We have obtained the all protein (~88,000) from SWISSPROT databases release 67 and the overlapping peptides of length nine have been generated for all proteins. For example, a protein of length n will have (n+1 – 9) overlapping peptides.
ii)Score of all natural 9-mer peptides have been calculated using quantitative matrix of that allele. These peptides have been sorted on the basis of score in descending order and top 1 % natural peptides have been obtained. The minimum score is determined from these selected peptides. This minimum score is called threshold score of at 1%. Similarly, threshold scores at 2%, 3% … 10% are calculated.
The step 1 and 2 is repeated for each allele, in order to calculate threshold score at different percent for each allele used in MMBPred.

Algorithm for Prediction of promiscuous Mutated MHC Binders


Random mutation: The following strategy is used to obtain mutated peptides with one random mutation at any position of peptide frame.
Firstly, the first position of peptide frame is mutated by all 20 amino acids while keeping all other amino acids conserved.
Secondly, the second position of peptide frame is mutated by all 20 amino acids while keeping all other amino acids conserved.
In this way, all possible mutated peptides will be produced for each position by changing one position with all amino acids while keeping all other amino acids conserved.

  • Generation of peptides with two position-specific or random mutations

    Positional mutations: Each peptide frame will have two mutations at user-defined positions. If the user wants to obtain peptides with mutation at first and ninth position of "ILKEPVHGV".
    Mutated peptides=N1LKEPVHGN9
    The N 1 and N9 can be any of the twenty amino acids. The mutated peptides will have all possible combinations of 20 amino acids at First and Ninth positions.

    Random mutations:-Random mutation corresponds to the mutations at any two randomly selected positions. e.g. The randomly mutated peptides can be obtained from "ILKEPVHGV" by taking into consideration all the possible combinations of twenty amino acids at any two positions (positions 1 & 2, 1 & 3, and so on)

  • Generation of peptides with three position-specific mutation

    Positional mutations: - These have three amino acids mutated in each peptide frame. All three user defined positions are substituted by all possible combinations of twenty amino acids. Thus, each peptide frame yields 6840 triple mutated peptides by all possible combinations. e.g for mutations at three specific positions (say 1, 2 & 9) from “ILKEPVHGV”, the mutated peptide will have all the possible combinations of twenty amino acids at first, second and ninth positions.

    TOP
  • Calculation of Peptide Score: Stepwise:
    1. The initial score for each peptide is set at 0.000.
    2. The program extracts the value of coefficient for a particular amino acid at particular position. The coefficient values are precalculated and stored in the tables of 20 X 9 known as quantitative matrices.


    e.g.
    Calculation of score for "ILKEYVHGV" by using HLA-A*0201 table.

    Amino acid/Position
    P1
    P2
    P3
    P4
    P5
    P6
    P7
    P8
    P9
    A
                     
    C
                     
    D
                     
    E
          0.56          
    F
                     
    G
                 
    -0.58
     
    H
               
    0.40
       
    I
    0.00
                   
    K
        -1.14            
    L
      6.31              
    M
                     
    N
                     
    P
                     
    Q
                     
    R
                     
    S
                     
    T
                     
    V
             
    0.58
        6.28
    W
                     
    Y
            0.68        
    TOP
  • The peptide score is obtained as a result of summation of each coefficient value.

  •    Initial score=0.000
       peptide score=I[1]+L[2]+K[3]+E[4]+Y[5]+V[6]+H[7]+G[8]+V[9].
       if peptide score  <  threshold score =Predicted Non binder
       if peptide score  >  threshold score=Predicted Binder
       
     The predicted mutated binders having score greater than native peptide frames are        considered as better mutated binders.


  • Prediction of Promiscuous mutated MHC binders The program examine the peptide having high binding affinity for many MHC alleles.The value of field "Peptide binding with[1-46] MHC allele is used as cutoff for selection of promiscious MHC binders.
    e.g.
    if the value of this field is "7".
    The result will display all peptide binding with 7 or more than seven MHC alleles.
  • TOP
    Algorithm for Epitope enhancement


    Here, the peptides frames are not mutated randomly by all amino acids at specific position as was the case of mutated binders prediction. But the mutation will be made by only those amino acids having highest favorable effect on peptide-MHC interaction. Suppose the user wants to enhance the binding affinity of "ILKEPVHGV" for specific MHC allele by mutating first & second positions while keeping all other positions conserved, then the first two positions are mutated by the amino acids having most favorable effect on MHC binding affinity of peptides. The amino acids having highest coefficient value at a particular position for specific MHC allele are considered as the most favorable amino acids.
    Enhanced epitope= N1N 2 KEPVHGV
    Where N1 and N2 are the amino acids having highest positive coefficient value at first and second positions respectively . The user can change any number of amino acids. The enhanced epitope with highest binding affinity for specific MHC allele is obtained if all the positions of the peptide are mutated.
    Note:The extraction of the overlapping peptide frame s from native antigenic protein and calculation of score for each mutated peptide will follwo the the procedure similiar to that is used in case of promiscious mutated binders prediction.

    TOP