1. Immunological Bioinformatics

2. The Immunological Bioinformatics group
Collaborators
IMMI, University of Copenhagen
Søren Buus MHC binding
Mogens H Claesson Elispot Assay
La Jolla Institute of Allergy and Infectious Diseases
Allesandro Sette Epitope database
Bjoern Peters
Leiden University Medical Center
Tom Ottenhoff Tuberculosis
Michel Klein
Ganymed
Ugur Sahin Genetic library
University of Tubingen
Stefan Stevanovic MHC ligands
INSERM
Peter van Endert Tap binding
University of Mainz
Hansjörg Schild Proteasome
Schafer-Nielsen
Claus Schafer-Nielsen Peptide synthesis
ImmunoGrid
Elda Rossi Simulation of the
Vladimir Brusic Immune system
University of Utrectht
Can Kesmir Ideas
Immunological Bioinformatics group, CBS, Technical University of Denmark (
Ole Lund, Group Leader
Morten Nielsen, Associate Professor
Claus Lundegaard , Associate Professor
Jean Vennestrøm, post doc.
Thomas Blicher (50%), post doc.
Mette Voldby Larsen, PhD student
Pernille Haste Andersen, PhD student
Sune Frankild, PhD student
Sheila Tang, PhD student
Thomas Rask (50%), PhD student
Nicolas Rapin , PhD student
Ilka Hoff , PhD student
Jorid Sørli, PhD student
Hao Zhang, PhD student
MSc students

3. Figure 1-20
Administration of a substance to a person with the purpose of preventing a disease
Traditionally composed of a killed or weakened micro organism
Vaccination works by creating a type of immune response that enables the memory cells to later respond to a similar organism before it can cause disease

4. Effectiveness of vaccines
1958 start of small pox eradication program

5. The Immune System
The innate immune system
The adaptive immune system

6. The innate immune system
Unspecific
Antigen independent
Immediate response
No training/selection hence no memory
Pathogen independent (but response might be pathogen type dependent)

7. The adaptive immune system
Pathogen specific
Humoral
Cellular
Parasite
Bacteria
Virus

8. Adaptive immune response
Signal induced
Pathogens
Antigens
Epitopes
B Cell
T Cell

9. Humoral immunity
Cartoon by Eric Reits

10. Antibody - Antigen interaction
The antibody recognizes structural properties of the surface of the antigen
Paratope
Fab
Epitope
Antibody

11. Cellular Immunity

12. MHC class I with peptide
Anchor positions

13. HLA specificity clustering
B0702

14. Prediction of HLA binding specificity Historical overview
Simple Motifs
Allowed/non allowed amino acids
Extended motifs
Amino acid preferences (SYFPEITHI)
Anchor/Preferred/other amino acids
Hidden Markov models
Peptide statistics from sequence alignment
SVMs and neural networks
Can take sequence correlations into account

15. Sequence information
SLLPAIVEL YLLPAIVHI TLWVDPYEV GLVPFLVSV KLLEPVLLL LLDVPTAAV LLDVPTAAV LLDVPTAAV
LLDVPTAAV VLFRGGPRG MVDGTLLLL YMNGTMSQV MLLSVPLLL SLLGLLVEV ALLPPINIL TLIKIQHTL
HLIDYLVTS ILAPPVVKL ALFPQLVIL GILGFVFTL STNRQSGRQ GLDVLTAKV RILGAVAKV QVCERIPTI
ILFGHENRV ILMEHIHKL ILDQKINEV SLAGGIIGV LLIENVASL FLLWATAEA SLPDFGISY KKREEAPSL
LERPGGNEI ALSNLEVKL ALNELLQHV DLERKVESL FLGENISNF ALSDHHIYL GLSEFTEYL STAPPAHGV
PLDGEYFTL GVLVGVALI RTLDKVLEV HLSTAFARV RLDSYVRSL YMNGTMSQV GILGFVFTL ILKEPVHGV
ILGFVFTLT LLFGYPVYV GLSPTVWLS WLSLLVPFV FLPSDFFPS CLGGLLTMV FIAGNSAYE KLGEFYNQM
KLVALGINA DLMGYIPLV RLVTLKDIV MLLAVLYCL AAGIGILTV YLEPGPVTA LLDGTATLR ITDQVPFSV
KTWGQYWQV TITDQVPFS AFHHVAREL YLNKIQNSL MMRKLAILS AIMDKNIIL IMDKNIILK SMVGNWAKV
SLLAPGAKQ KIFGSLAFL ELVSEFSRM KLTPLCVTL VLYRYGSFS YIGEVLVSV CINGVCWTV VMNILLQYV
ILTVILGVL KVLEYVIKV FLWGPRALV GLSRYVARL FLLTRILTI HLGNVKYLV GIAGGLALL GLQDCTMLV
TGAPVTYST VIYQYMDDL VLPDVFIRC VLPDVFIRC AVGIGIAVV LVVLGLLAV ALGLGLLPV GIGIGVLAA
GAGIGVAVL IAGIGILAI LIVIGILIL LAGIGLIAA VDGIGILTI GAGIGVLTA AAGIGIIQI QAGIGILLA
KARDPHSGH KACDPHSGH ACDPHSGHF SLYNTVATL RGPGRAFVT NLVPMVATV GLHCYEQLV PLKQHFQIV
AVFDRKSDA LLDFVRFMG VLVKSPNHV GLAPPQHLI LLGRNSFEV PLTFGWCYK VLEWRFDSR TLNAWVKVV
GLCTLVAML FIDSYICQV IISAVVGIL VMAGVGSPY LLWTLVVLL SVRDRLARL LLMDCSGSI CLTSTVQLV
VLHDDLLEA LMWITQCFL SLLMWITQC QLSLLMWIT LLGATCMFV RLTRFLSRV YMDGTMSQV FLTPKKLQC
ISNDVCAQV VKTDGNPPE SVYDFFVWL FLYGALLLA VLFSSDFRI LMWAKIGPV SLLLELEEV SLSRFSWGA
YTAFTIPSI RLMKQDFSV RLPRIFCSC FLWGPRAYA RLLQETELV SLFEGIDFY SLDQSVVEL RLNMFTPYI
NMFTPYIGV LMIIPLINV TLFIGSHVV SLVIVTTFV VLQWASLAV ILAKFLHWL STAPPHVNV LLLLTVLTV
VVLGVVFGI ILHNGAYSL MIMVKCWMI MLGTHTMEV MLGTHTMEV SLADTNSLA LLWAARPRL GVALQTMKQ
GLYDGMEHL KMVELVHFL YLQLVFGIE MLMAQEALA LMAQEALAF VYDGREHTV YLSGANLNL RMFPNAPYL
EAAGIGILT TLDSQVMSL STPPPGTRV KVAELVHFL IMIGVLVGV ALCRWGLLL LLFAGVQCQ VLLCESTAV
YLSTAFARV YLLEMLWRL SLDDYNHLV RTLDKVLEV GLPVEYLQV KLIANNTRV FIYAGSLSA KLVANNTRL
FLDEFMEGV ALQPGTALL VLDGLDVLL SLYSFPEPE ALYVDSLFF SLLQHLIGL ELTLGEFLK MINAYLDKL
AAGIGILTV FLPSDFFPS SVRDRLARL SLREWLLRI LLSAWILTA AAGIGILTV AVPDEIPPL FAYDGKDYI
AAGIGILTV FLPSDFFPS AAGIGILTV FLPSDFFPS AAGIGILTV FLWGPRALV ETVSEQSNV ITLWQRPLV

16. Scoring a sequence to a weight matrix
Score sequences to weight matrix by looking up and adding L values from the matrix
A R N D C Q E G H I L K M F P S T W Y V
Which peptide is most likely to bind?
Which peptide second?
RLLDDTPEV
GLLGNVSTV
ALAKAAAAL
11.9
14.7
4.3
84nM
23nM
309nM

17. Example from real life 10 peptides from MHCpep database
Bind to the MHC complex
Relevant for immune system recognition
Estimate sequence motif and weight matrix
Evaluate motif “correctness” on 528 peptides
ALAKAAAAM
ALAKAAAAN
ALAKAAAAR
ALAKAAAAT
ALAKAAAAV
GMNERPILT
GILGFVFTM
TLNAWVKVV
KLNEPVLLL
AVVPFIVSV

18. Prediction accuracy Measured affinity Prediction score
Pearson correlation 0.45
Measured affinity
Prediction score

19. Predictive performance

20. Higher order sequence correlations
Neural networks can learn higher order correlations!
What does this mean?
Say that the peptide needs one and only one large amino acid in the positions P3 and P4 to fill the binding cleft
How would you formulate this to test if a peptide can bind?
S S => 0
L S => 1
S L => 1
L L => 0
No linear function can learn this (XOR) pattern

21. Mutual information
313 binding peptides
313 random peptides

22. Sequence encoding (continued)
Sparse encoding V: L:
V.L=0 (unrelated)
Blosum encoding V: L: R:
V.L = (highly related)
V.R = (close to unrelated)

23. Evaluation of prediction accuracy

24. Network ensembles
No one single network with a particular architecture and sequence encoding scheme, will constantly perform the best
Also for Neural network predictions will enlightened despotism fail
For some peptides, BLOSUM encoding with a four neuron hidden layer can best predict the peptide/MHC binding, for other peptides a sparse encoded network with zero hidden neurons performs the best
Wisdom of the Crowd
Never use just one neural network
Use Network ensembles

25. Evaluation of prediction accuracy
ENS: Ensemble of neural networks trained using sparse,
Blosum, and hidden Markov model sequence encoding

26. IEDB + more proprietary data
NetMHC-3.0 update
IEDB + more proprietary data
Higher accuracy for existing ANNs
More Human alleles
Non human alleles (Mice + Primates)
Prediction of 8mer binding peptides for some alleles
Prediction of 10- and 11mer peptides for all alleles
Outputs to spread sheet

27. Prediction of 10- and 11mers using 9mer prediction tools
Approach:
For each peptide of length L create 6 pseudo peptides deleting a sliding window of L- 9 always keeping pos. 1,2,3, and 9
Example:
MLPQWESNTL = MLPWESNTL
MLPQESNTL
MLPQWSNTL
MLPQWENTL
MLPQWESTL
MLPQWESNL

28. Prediction of 10- and 11mers using 9mer prediction tools

29. Prediction of 10- and 11mers using 9mer prediction tools
Final prediction = average of the 6 log scores:
( )/6
= 0.505
Affinity:
Exp(log(50000)*( )) = nM

30. Prediction using ANN trained on 10mer peptides

31. Prediction of 10- and 11mers using 9mer prediction tools

32. Cellular Immunity

33. Proteasome specificity
Low polymorphism
Constitutive & Immuno-proteasome
Evolutionary conserved
Stochastic and low specificity
Only 70-80% of the cleavage sites are reproduced in repeated experiments

34. Proteasome specificity
NetChop is one of the best available cleavage method

35. Predicting TAP affinity
9 meric peptides
>9 meric
ILRGTSFVYV
= -0.74
Peters et el., JI, 171: 1741.

36. Integration?
Integrating all three steps (protesaomal cleavage, TAP transport and MHC binding) should lead to improved identification of peptides capable of eliciting CTL responses

37. Identifying CTL epitopes
HLA affinity
Proteasomal cleavage
TAP affinity
1 EBN3_EBV YQAYSSWMY
2 EBN3_EBV QSDETATSH
3 EBN3_EBV PVSPAVNQY
4 EBN3_EBV AYSSWMYSY
5 EBN3_EBV LAAGWPMGY
6 EBN3_EBV IVQSCNPRY
7 EBN3_EBV FLQRTDLSY
8 EBN3_EBV YTDHQTTPT
9 EBN3_EBV GTDVVQHQL
...

40. Large scale method validation
HIV A3 epitope predictions

41. Sylvester-Hvid et al, Tissue Antigens. 2004
Case I: SARS
Sylvester-Hvid et al, Tissue Antigens. 2004

42. 75% of predicted peptides were binding with an IC50 <500 nM
Sars virus HLA ligands
75% of predicted peptides were binding with an IC50 <500 nM

43. Case II: Discovery of conserved Class I epitopes in Human Influenza Virus H1N1
Wang et al., Vaccine 2007

44. Pox Strategy

45. Influenza
We selected the Influenza peptides with the top 15 combined scores with conservation p9 > 70% for each pf the 12 supertypes.
180 peptides selected
167 tested for binding and CTL response
89 (53%) of the influenza peptides tested have an affinity better than 500nM

46. Donors 35 normal healthy blood donors 35-65 years old
Expected to have had influenza more than 3 times
HLA typed by SBT for HLA A and B

47. ELISPOT assay
Measure number of white blood cells that in vitro produce interferon-g in response to a peptide
A positive result means that the immune system have earlier reacted to the peptide (during a response of a vaccine/natural infection)
FLDVMESM
FLDVMESM
FLDVMESM
FLDVMESM
FLDVMESM
FLDVMESM
Two spots

48. Peptides positive in ELISPOT assay

49. Conservation of epitopes
Number of 9mers 100% conserved:
10/12 conserved in Influenza A virus (A/Goose/Guangdong/1/96(H5N1))
11/12 conserved in Influenza A virus (A/chicken/Jilin/9/2004(H5N1))

50. Repeat until the desired number of peptides is selected
EpiSelect
Select peptide with maximal coverage
Top Scoring Peptides
Genotype 1
Genotype 2
Select peptide with maximal coverage preferring uncovered strains
Genotype 3
Genotype 4
Genotype 5
Genotype 6
Select peptide with maximal coverage preferring lowest covered strains
Repeat until the desired number of peptides is selected

51. HCV Results - B7 Genotype 1 Genotype 2 Genotype 3 Genotype 4
Peptides
Genome
Coverage
Predicted
affinity (nM)
Peptide
Genotype 1 4
QPRGRRQPI 5 5
Genotype 2 3
SPRGSRPSW 43 4 2
Genotype 3
DPRRRSRNL* 66 3
Genotype 4 3
RARAVRAKL 6 3
Genotype 5 3
31 patients with different ethnicity infected with different subtypes.
30 out of 31 had response towards at least one peptide.
116 of our 185 peptides (62%) did induce a response in at least one patient.
21 of the recognized peptides induced a response in >4 patients (13% of the study subjects).
TPAETTVRL* 38 3
Genotype 6 3
* Verified B7 supertype restricted CD8+ epitope in the Los Alamos HCV epitope database

52. Selection of epitopes covering host (HLA) and pathogen variability
Ongoing work
Selection of epitopes covering host (HLA) and pathogen variability
Selection of diagnostic peptides in TB
Predict cross reactivity (T and B cell)
Applications in epitope prediction, autoimmune diseases, transplantation
Virulence factor discovery by comparative genomics
Function-antigenecity studies
Bioinformatics immune system simulation