Presentation is loading. Please wait.

Presentation is loading. Please wait.

CENTER FOR BIOLOGICAL SEQUENCE ANALYSIS Technical University of Denmark - DTU Department of systems biology B cell epitopes and predictions.

Similar presentations


Presentation on theme: "CENTER FOR BIOLOGICAL SEQUENCE ANALYSIS Technical University of Denmark - DTU Department of systems biology B cell epitopes and predictions."— Presentation transcript:

1 CENTER FOR BIOLOGICAL SEQUENCE ANALYSIS Technical University of Denmark - DTU Department of systems biology B cell epitopes and predictions

2 Technical University of Denmark - DTU Department of systems biology ECCB/ISMB Immunological Bioinformatics Tutorial Outline What is a B-cell epitope? How can you predict B-cell epitopes?

3 Technical University of Denmark - DTU Department of systems biology ECCB/ISMB Immunological Bioinformatics Tutorial What is a B-cell epitope? Antibody Fab fragment B-cell epitopes: Accessible structural feature of a pathogen molecule. Antibodies are developed to bind the epitope specifically using the complementary determining regions (CDRs).

4 Technical University of Denmark - DTU Department of systems biology ECCB/ISMB Immunological Bioinformatics Tutorial The binding interactions Salt bridges Hydrogen bonds Hydrophobic interactions Van der Waals forces Binding strength

5 Technical University of Denmark - DTU Department of systems biology ECCB/ISMB Immunological Bioinformatics Tutorial The binding interactions Salt bridges Hydrogen bonds Hydrophobic interactions Van der Waals forces Binding strength

6 Technical University of Denmark - DTU Department of systems biology ECCB/ISMB Immunological Bioinformatics Tutorial The binding interaction However spatial composition matters!!!

7 Technical University of Denmark - DTU Department of systems biology ECCB/ISMB Immunological Bioinformatics Tutorial The binding interaction

8 Technical University of Denmark - DTU Department of systems biology ECCB/ISMB Immunological Bioinformatics Tutorial B-cell epitope classification Linear epitopes One segment of the amino acid chain Discontinuous epitope (with linear determinant) Discontinuous epitope Several small segments brought into proximity by the protein fold B-cell epitope: structural feature of a molecule or pathogen, accessible and recognizable by B-cell receptors and antibodies

9 Technical University of Denmark - DTU Department of systems biology ECCB/ISMB Immunological Bioinformatics Tutorial B-cell epitope annotation Linear epitopes: Chop sequence into small pieces and measure binding to antibody Discontinuous epitopes: Measure binding of whole protein to antibody The best annotation method : X-ray crystal structure of the antibody-epitope complex

10 Technical University of Denmark - DTU Department of systems biology ECCB/ISMB Immunological Bioinformatics Tutorial B-cell epitope annotation Linear epitopes: Chop sequence into small pieces and measure binding to antibody Discontinuous epitopes: Measure binding of whole protein to antibody The best annotation method : X-ray crystal structure of the antibody-epitope complex 10% 90%

11 Technical University of Denmark - DTU Department of systems biology ECCB/ISMB Immunological Bioinformatics Tutorial B-cell epitope annotation No epitope is purely linear Epitopes contains linear determinants of 5 or more residues Longest linear stretch in epitope

12 Technical University of Denmark - DTU Department of systems biology ECCB/ISMB Immunological Bioinformatics Tutorial B-cell epitope data bases Databases: IEDB, Los Alamos HIV database, Protein Data Bank, AntiJen, BciPep Large amount of data available for linear epitopes Few data available for discontinuous

13 CENTER FOR BIOLOGICAL SEQUENCE ANALYSIS Technical University of Denmark - DTU Department of systems biology B cell epitope prediction

14 CENTER FOR BIOLOGICAL SEQUENCE ANALYSIS Technical University of Denmark - DTU Department of systems biology The immune system – prediction tools Cytotoxic T cell epitope: (A ROC ~ 0.9) Will a given peptide bind to a given MHC class I molecule Helper T cell Epitope (A ROC ~ 0.85) Will a part of a peptide bind to a given MHC II molecule B cell epitope (A ROC ~ 0.74) Will a given part of a protein bind to one of the billions of different B Cell receptors

15 CENTER FOR BIOLOGICAL SEQUENCE ANALYSIS Technical University of Denmark - DTU Department of systems biology

16 CENTER FOR BIOLOGICAL SEQUENCE ANALYSIS Technical University of Denmark - DTU Department of systems biology B-cell – prediction tools Sequence based prediction tools –Predominantly predicts linear epitopes Structure based epitopes – Predicts Conformational epitopes

17 CENTER FOR BIOLOGICAL SEQUENCE ANALYSIS Technical University of Denmark - DTU Department of systems biology Watch out for overfitting

18 Technical University of Denmark - DTU Department of systems biology ECCB/ISMB Immunological Bioinformatics Tutorial Sequence-based methods for prediction of linear epitopes TSQDLSVFPLASCCKDNIASTSVTLGCLVTGY LPMSTTVTWDTGSLNKNVTTFPTTFHETYGL HSIVSQVTASGKWAKQRFTCSVAHAESTAINK TFSACALNFIPPTVKLFHSSCNPVGDTHTTIQL LCLISGYVPGDMEVIWLVDGQKATNIFPYTAP GTKEGNVTSTHSELNITQGEWVSQKTYTCQV TYQGFTFKDEARKCSESDPRGVTSYLSPPSP L Input TSQDLSVFPLASCCKDNIASTSVTLGCLVTGY LPMSTTVTWDTGSLNKNVTTFPTTFHETYGL HSIVSQVTASGKWAKQRFTCSVAHAESTAINK TFSACALNFIPPTVKLFHSSCNPVGDTHTTIQL LCLISGYVPGDMEVIWLVDGQKATNIFPYTAP GTKEGNVTSTHSELNITQGEWVSQKTYTCQV TYQGFTFKDEARKCSESDPRGVTSYLSPPSP L Output

19 Technical University of Denmark - DTU Department of systems biology ECCB/ISMB Immunological Bioinformatics Tutorial Sequence-based methods for prediction of linear epitopes Protein hydrophobicity – hydrophilicity algorithms Parker, Fauchere, Janin, Kyte and Doolittle, Manavalan Sweet and Eisenberg, Goldman, Engelman and Steitz (GES), von Heijne Protein flexibility prediction algorithm Karplus and Schulz Protein secondary structure prediction algorithms PsiPred (D. Jones)

20 Technical University of Denmark - DTU Department of systems biology ECCB/ISMB Immunological Bioinformatics Tutorial Propensity scales: The principle The Parker hydrophilicity scale Derived from experimental data D2.46 E1.86 N1.64 S1.50 Q1.37 G1.28 K1.26 T1.15 R0.87 P0.30 H0.30 C0.11 A0.03 Y-0.78 V-1.27 M-1.41 I-2.45 F-2.78 L-2.87 W-3.00 Hydrophilicity

21 Technical University of Denmark - DTU Department of systems biology ECCB/ISMB Immunological Bioinformatics Tutorial ….LISTFVDEKRPGSDIVEDLILKDENKTTVI…. ( )/7 = 0.39 Prediction scores: Epitope Propensity scales: The principle

22 Technical University of Denmark - DTU Department of systems biology ECCB/ISMB Immunological Bioinformatics Tutorial Evaluation of performance

23 Technical University of Denmark - DTU Department of systems biology ECCB/ISMB Immunological Bioinformatics Tutorial Turn prediction and B-cell epitopes Pellequer found that 50% of the epitopes in a data set of 11 proteins were located in turns Turn propensity scales for each position in the turn were used for epitope prediction. Pellequer et al., Immunology letters,

24 Technical University of Denmark - DTU Department of systems biology ECCB/ISMB Immunological Bioinformatics Tutorial Blythe and Flower 2005 Extensive evaluation of propensity scales for epitope prediction Conclusion: – Basically all the classical scales perform close to random! – Other methods must be used for epitope prediction

25 Technical University of Denmark - DTU Department of systems biology ECCB/ISMB Immunological Bioinformatics Tutorial BepiPred Parker hydrophilicity scale PSSM PSSM based on linear epitopes extracted from the AntiJen database Combination of the Parker prediction scores and PSSM leads to prediction score Tested on the Pellequer dataset and epitopes in the HIV Los Alamos database

26 Technical University of Denmark - DTU Department of systems biology ECCB/ISMB Immunological Bioinformatics Tutorial PSSM ….LISTFVDEKRPGSDIVEDLILKDENKTTVI… = 6.75 Prediction value A R N D C Q E G H I L K M F P S T W Y V S I

27 Technical University of Denmark - DTU Department of systems biology ECCB/ISMB Immunological Bioinformatics Tutorial ROC evaluation Evaluation on HIV Los Alamos data set

28 Technical University of Denmark - DTU Department of systems biology ECCB/ISMB Immunological Bioinformatics Tutorial BepiPred performance Pellequer data set: – LevittAROC = 0.66 – ParkerAROC = 0.65 – BepiPredAROC = 0.68 HIV Los Alamos data set – LevittAROC = 0.57 – ParkerAROC = 0.59 – BepiPredAROC = 0.60

29 Technical University of Denmark - DTU Department of systems biology ECCB/ISMB Immunological Bioinformatics Tutorial BepiPred BepiPred conclusion: On both of the evaluation data sets, Bepipred was shown to perform better Still the AROC value is low compared to T-cell epitope prediction tools! Bepipred is available as a webserver:

30 Technical University of Denmark - DTU Department of systems biology ECCB/ISMB Immunological Bioinformatics Tutorial Prediction of linear epitopes Con only ~10% of epitopes can be classified as “linear” weakly immunogenic in most cases most epitope peptides do not provide antigen-neutralizing immunity in many cases represent hypervariable regions Pro easily predicted computationally easily identified experimentally immunodominant epitopes in many cases do not need 3D structural information easy to produce and check binding activity experimentally

31 Technical University of Denmark - DTU Department of systems biology ECCB/ISMB Immunological Bioinformatics Tutorial Sequence based prediction methods Linear methods for prediction of B cell epitopes have low performances The problem is analogous to the problems of representing the surface of the earth on a two-dimensional map Reduction of the dimensions leads to distortions of scales, directions, distances The world of B-cell epitopes is 3 dimensional and therefore more sophisticated methods must be developed

32 Technical University of Denmark - DTU Department of systems biology ECCB/ISMB Immunological Bioinformatics Tutorial So what is more sophisticated? Use of the three dimensional structure of the pathogen protein Analyze the structure to find surface exposed regions Additional use of information about conformational changes, glycosylation and trans-membrane helices

33 Technical University of Denmark - DTU Department of systems biology ECCB/ISMB Immunological Bioinformatics Tutorial Sources of three-dimensional structures Experimental determination X-ray crystallography NMR spectroscopy Both methods are time consuming and not easily done in a larger scale Structure prediction Homology modeling Fold recognition Less time consuming, but there is a possibility of incorrect predictions, specially in loop regions

34 Technical University of Denmark - DTU Department of systems biology ECCB/ISMB Immunological Bioinformatics Tutorial Protein structure prediction methods Homology/comparative modeling >25% sequence identity (seq 2 seq alignment) Fold-recognition <25% sequence identity (Psi-blast search/ PSSM 2 seq alignment) Ab initio structure prediction 0% sequence identity

35 Technical University of Denmark - DTU Department of systems biology ECCB/ISMB Immunological Bioinformatics Tutorial What does antibodies recognize in a protein?

36 Technical University of Denmark - DTU Department of systems biology ECCB/ISMB Immunological Bioinformatics Tutorial The binding interaction

37 Technical University of Denmark - DTU Department of systems biology ECCB/ISMB Immunological Bioinformatics Tutorial The binding interaction However spatial composition matters!!!

38 Technical University of Denmark - DTU Department of systems biology ECCB/ISMB Immunological Bioinformatics Tutorial What does antibodies recognize in a protein? probe Antibody Fab fragment Protrusion index Surface Exposed Hydrophobic region

39 Technical University of Denmark - DTU Department of systems biology ECCB/ISMB Immunological Bioinformatics Tutorial Structure base prediction methods

40 Technical University of Denmark - DTU Department of systems biology ECCB/ISMB Immunological Bioinformatics Tutorial DiscoTope Prediction of residues in discontinuous B cell epitopes using protein 3D structures Pernille Haste Andersen, Morten Nielsen and Ole Lund, Protein Science 2006

41 Præsentationens navn41DTU Systembiologi, Danmarks Tekniske Universitet Predicting B-cell epitopes

42 Præsentationens navn42DTU Systembiologi, Danmarks Tekniske Universitet The DiscoTope method

43 Præsentationens navn43DTU Systembiologi, Danmarks Tekniske Universitet Some amino acids are preferred and disliked in the epitope The DiscoTope method Epitope amino acid preferences Kringelum et al, 2011, manuscript in preparation

44 Præsentationens navn44DTU Systembiologi, Danmarks Tekniske Universitet Some amino acids are preferred and disliked in the epitope Epitopes reside on the surface of the protein The DiscoTope method Kringelum et al, 2011, manuscript in preparation

45 Præsentationens navn45DTU Systembiologi, Danmarks Tekniske Universitet Predictions score for each residue are calculated by summing the epitope likelihood (propensity) of surrounding residues and subtracting the neighbor count The DiscoTope method Andersen et al., 2006

46 Præsentationens navn46DTU Systembiologi, Danmarks Tekniske Universitet Predictions score for each residue are calculated by summing the epitope likelihood (propensity) of surrounding residues and subtracting the neighbor count The DiscoTope method Performance: A roc = Andersen et al., 2006 On a dataset of 75 antigen- antibody complexes divided in 25 proteins

47 Præsentationens navn47DTU Systembiologi, Danmarks Tekniske Universitet Predictions score for each residue are calculated by summing the epitope likelihood (propensity) of surrounding residues and subtracting the neighbor count The DiscoTope method Performance: A roc = Andersen et al., 2006 On a dataset of 75 antigen- antibody complexes divided in 25 proteins

48 Præsentationens navn48DTU Systembiologi, Danmarks Tekniske Universitet Uneven spatial distribution of amino acid in epitopes However position matters Kringelum et al, 2012, manuscript in review

49 17/04/2008Presentation name49DTU Sytems Biology, Technical University of Denmark Propensity score function Amino acid log-odds score Weight factor

50 17/04/2008Presentation name50DTU Sytems Biology, Technical University of Denmark PS(THR256) Propensity score function

51 17/04/2008Presentation name51DTU Sytems Biology, Technical University of Denmark 1)Identify Naighbor residues within 21.6Å (C α -C α ) PS(THR256) Propensity score function Radius = 21.6Å

52 17/04/2008Presentation name52DTU Sytems Biology, Technical University of Denmark PS(THR256) Propensity score function 1)Identify Naighbor residues within 21.6Å (C α -C α ) Radius = 21.6Å

53 17/04/2008Presentation name53DTU Sytems Biology, Technical University of Denmark PS(THR256) Propensity score function 1)Identify Naighbor residues within 21.6Å (C α -C α ) 2)Calculate summed propensity score: ls(THR) = β 256 = 0.8*(1-0.0/21.6)+0.2 = 1.0 ls(THR)*β 256 = -0.23*1.0 = d 256 = 0.0Å

54 17/04/2008Presentation name54DTU Sytems Biology, Technical University of Denmark PS(THR256) Propensity score function 1)Identify Naighbor residues within 21.6Å (C α -C α ) 2)Calculate summed propensity score: d 255 = 3.8Å ls(THR) = β 256 = 0.8*(1-0.0/21.6)+0.2 = 1.0 ls(THR)*β 256 = -0.23*1.0 = ls(ASP) = 2.5 β 255 =0.8*(1-3.8/21.6)+0.2 = 0.86 ls(THR)*β 255 = 2.5*0.86 = 2.15

55 17/04/2008Presentation name55DTU Sytems Biology, Technical University of Denmark PS(THR256) Propensity score function 1)Identify Naighbor residues within 21.6Å (C α -C α ) 2)Calculate summed propensity score: ls(THR) = β 256 = 0.8*(1-0.0/21.6)+0.2 = 1.0 ls(THR)*β 256 = -0.23*1.0 = ls(ASP) = 2.5 β 255 =0.8*(1-3.8/21.6)+0.2 = 0.86 ls(ASP)*β 255 = 2.5*0.86 = 2.15 ls(THR) = β 254 =0.8*(1-6.1/21.6)+0.2 = 0.77 ls(THR)*β 254 = -0.23*0.77 = d 254 = 6.1Å

56 17/04/2008Presentation name56DTU Sytems Biology, Technical University of Denmark PS(THR256) Propensity score function 1)Identify Naighbor residues within 21.6Å (C α -C α ) 2)Calculate summed propensity score: ls(THR) = β 256 = 0.8*(1-0.0/21.6)+0.2 = 1.0 ls(THR)*β 256 = -0.23*1.0 = ls(ASP) = 2.5 β 255 =0.8*(1-3.8/21.6)+0.2 = 0.86 ls(ASP)*β 255 = 2.5*0.86 = 2.15 ls(THR) = β 254 =0.8*(1-6.1/21.6)+0.2 = 0.77 ls(THR)*β 254 = -0.23*0.77 = … …………. ls(PRO) = 1.2 β 254 =0.8*(1-20.6/21.6)+0.2 = 0.24 ls(PRO)*β 254 = 1.2*0.24 = 0.29 d 247 = 20.6Å

57 17/04/2008Presentation name57DTU Sytems Biology, Technical University of Denmark PS(THR256) Propensity score function 1)Identify Naighbor residues within 21.6Å (C α -C α ) 2)Calculate summed propensity score: ls(THR) = β 256 = 0.8*(1-0.0/21.6)+0.2 = 1.0 ls(THR)*β 256 = -0.23*1.0 = ls(ASP) = 2.5 β 255 =0.8*(1-3.8/21.6)+0.2 = 0.86 ls(ASP)*β 255 = 2.5*0.86 = 2.15 ls(THR) = β 254 =0.8*(1-6.1/21.6)+0.2 = 0.77 ls(THR)*β 254 = -0.23*0.77 = … …………. ls(PRO) = 1.2 β 254 =0.8*(1-20.6/21.6)+0.2 = 0.24 ls(PRO)*β 254 = 1.2*0.24 = 0.29 Summation of scores PS(THR256) = 0.39

58 17/04/2008Presentation name58DTU Sytems Biology, Technical University of Denmark HS(THR256) Half-sphere exposure 1)Create the upper half-sphere, which is the half-sphere where the residue side-chain is pointing

59 17/04/2008Presentation name59DTU Sytems Biology, Technical University of Denmark HS(THR256) Half-sphere exposure 1)Create the upper half-sphere, which is the half-sphere where the residue side-chain is pointing 2)Count neighbor residues within the half-sphere (nr of C α -atoms) 3)As high counts means highly buried the counts are multiplied by -1 = 5 HS(THR256) = -5

60 17/04/2008Presentation name60DTU Sytems Biology, Technical University of Denmark 1)Calculate Propensity score 2)Calculate half-sphere exposure 3)The final score is a weighted sum of the Propensity score and half-sphere exposure PS(THR256) = 0.39 HS(THR256) = -5 α = DS = (1-α)*PS + α*HS DS(THR256)= 0.885* *(- 5) DS(THR256)= DS(THR256) The DiscoTope 2.0 Score

61 Præsentationens navn61DTU Systembiologi, Danmarks Tekniske Universitet Performance and limitations Average AUC = 0.741

62 Præsentationens navn62DTU Systembiologi, Danmarks Tekniske Universitet Performance and limitations Glycosylated proteins

63 Præsentationens navn63DTU Systembiologi, Danmarks Tekniske Universitet Performance and limitations Gp120Hemagglutinine Glycosylation effects predictions

64 Præsentationens navn64DTU Systembiologi, Danmarks Tekniske Universitet Performance and limitations Small fragments (<120 residues) of larger biological units

65 Præsentationens navn65DTU Systembiologi, Danmarks Tekniske Universitet Inclusion of biological units enhance performance Performance and limitations Potassium Channel

66 Præsentationens navn66DTU Systembiologi, Danmarks Tekniske Universitet External Benchmark Dataset 52 antigen:antibody structures 33 homology groups Performance: AUC Performance and limitations

67 Præsentationens navn67DTU Systembiologi, Danmarks Tekniske Universitet Conclusions DiscoTope V2.0 outperforms similar methods High performance on 15/25 Medium performance on 7/25 Fail on 3/25 Inclusion of surface measures does only slightly enhance predictions Use the entire biological unit, when possible Small fragments (< 120 residues) have lower performance Glycosylation might course the prediction to fail Check for clash between predicted epitopes and glycosylation sites

68 Technical University of Denmark - DTU Department of systems biology ECCB/ISMB Immunological Bioinformatics Tutorial Rational vaccine design >PATHOGEN PROTEIN KVFGRCELAAAMKRHGLDNYRGY SLGNWVCAAKFESNF Rational Vaccine Design

69 Technical University of Denmark - DTU Department of systems biology ECCB/ISMB Immunological Bioinformatics Tutorial Rational B-cell epitope design Protein target choice Structural analysis of antigen Known structure or homology model Precise domain structure Physical annotation (flexibility, electrostatics, hydrophobicity) Functional annotation (sequence variations, active sites, binding sites, glycosylation sites, etc.) Known 3D structure Model

70 Technical University of Denmark - DTU Department of systems biology ECCB/ISMB Immunological Bioinformatics Tutorial Protein target choice Structural annotation Epitope prediction and ranking Rational B-cell epitope design Surface accessibility Protrusion index Conserved sequence Glycosylation status

71 Technical University of Denmark - DTU Department of systems biology ECCB/ISMB Immunological Bioinformatics Tutorial Rational B-cell epitope design Protein target choice Structural annotation Epitope prediction and ranking Optimal Epitope presentation Fold minimization, or Design of structural mimics Choice of carrier (conjugates, DNA plasmids, virus like particles) Multiple chain protein engineering

72 Technical University of Denmark - DTU Department of systems biology ECCB/ISMB Immunological Bioinformatics Tutorial Conclusions Rational vaccines can be designed to induce strong and epitope-specific B-cell responses Selection of protective B-cell epitopes involves structural, functional and immunogenic analysis of the pathogenic proteins When you can: Use protein structure for prediction Structural modeling tools are helpful in prediction of epitopes, design of epitope mimics and optimal epitope presentation


Download ppt "CENTER FOR BIOLOGICAL SEQUENCE ANALYSIS Technical University of Denmark - DTU Department of systems biology B cell epitopes and predictions."

Similar presentations


Ads by Google