1. Comparative Modelling Threading Baldomero Oliva Miguel UniversitatPompeuFabra

2. EVOLUTIONEVOLUTION

3. Diferences(proteins) Genetic information (DNA)

4. DNA secuence 3D

5. The Sanger Centre U.S. Genome Project ( 1990 ) Identify human genes Store the information in Data Banks Development of tools for analisis Venter, Patrinos & Collins 19981998 DOEDOE NHGRINHGRINIHNIH

6. Increase of sequences 10500000GenBank (DNA) 378152TrEMBL 92211Swiss Prot

7. Public Database Holdings:

8. 3D known (20%) Homologs (30%) Remotes (10%) Unknown function (20%) NO homology (20%) sec.3D

9. Knowing the structure helps to: Knowing the structure helps to: TO KNOW TO KNOW the function: annotation IMPROVE IMPROVE the function: genetic engineering INHIBIT INHIBIT the function: drugs

10. Structure Factory express & purify cristallize X-ray analises structure

11. Objetive Enough information Not accurate NO information 3D

12. How does modelling help? Structural Genomics X-ray

13. Can we predict protein structures from genome sequences?

14. Study of known structures Assigning structure to sequences with unknown structure

15. secondary Tertiary Met Val Leu Tyr Asp Ser Thr sequence quaternary super-secondary domain Caracterize the conformation Ligand docking

16. TrioseFosfateIsomerase Carboxypeptidase Glucanase Ribonuclease Mioglobin

17. The number of different protein folds is limited: New Folds Known Folds

18. The number of different protein folds is limited: [ last update: Oct 2001 ]

19. Nº sequences > Nº folds Structura 3ª Classification

20. UCL, Janet Thornton & Christine Orengo Class (C), Architecture(A), Topology(T), Homologous superfamily (H) Protein Structure Classification CATH - Protein Structure Classification [ http://www.biochem.ucl.ac.uk/bsm/cath_new/ ] SCOP - Structural Classification of Proteins MRC Cambridge (UK), Alexey Murzin, Brenner S. E., Hubbard T., Chothia C. created by manual inspection comprehensive description of the structural and evolutionary relationships [ http://scop.mrc-lmb.cam.ac.uk/scop/ ]

21. Class(C) derived from secondary structure content is assigned automatically Architecture(A) describes the gross orientation of secondary structures, independent of connectivity. Topology(T) clusters structures according to their topological connections and numbers of secondary structures Homologous superfamily (H)

22. TSS toxin 8.8% Id Remote Cholera toxin 80% Id Homolog SCOP Family Superfamily Fold Class Enterotoxin Aminoacyl tRNA synthetase 4.4% Id Analog

23. MNIFEMLRID EGLRLKIYKD TEGYYTIGIG HLLTKSPSLN AAKSELDKAI GRNCNGVITK DEAEKLFNQD VDAAVRGILR NAKLKPVYDS LDAVRRCALI NMVFQMGETG VAGFTNSLRM LQQKRWDEAA VNLAKSRWYN QTPNRAKRVI TTFRTGTWDA YKNL Can we predict protein structures ?

24. Homology modeling Idea: Extrapolation of the structure for a new (target) sequence from the known 3D-structures of related family members (templates).

25. Alignments Nº sequences > Nº folds Step 1 Selection of templates Step 2 Alignment with templates

26. . 0 20 40 60 80 100 050100150200250 identity Number of residues aligned Percentage sequence identity/similarity (B.Rost, Columbia, NewYork) Sequence identity implies structural similarity Don’t know region..... Sequence similarity implies structural similarity?

27. Hidropaticity Plot 2D Structure 2D prediction Nº sequences > Nº folds

28. 2D prediction (PHD)

29. Alignments Nº sequences > Nº folds Structure 3D Classification - Homologs - Remotes - Analogs Hidropaticity Plot 2D Structure 2D prediction

30. Comparative Modelling (homology) Fold is more conserved than sequence. Secondary structure is the most conserved structure Loops have the higher variability in structure.

31. Comparative modelling - Composer - Modeler SCR & VR SCR VR - Ring Closure - DB Search - Classification Alignments Nº sequences > Nº folds Structure 3D Classification - Homologs - Remotes - Analogs Hidropaticity Plot 2D Structure 2D prediction

32. MODEL Comparative modelling - Composer - Modeler SCR & VR SCR VR - Ring Closure - DB Search - Classification Alignments Nº sequences > Nº folds Structure 3D Classification - Homologs - Remotes - Analogs Hidropaticity Plot 2D Structure 2D prediction Step 3 Model Building

33. Fragment Extraction Rigid-body Assembly Loop Construction Spatial Constrains Optimisation Methods Satisfaction of Spatial Constrains

34. averaging core template backbone atoms (weighted by local sequence similarity with the target sequence) Leave non-conserved regions (loops) for later …. a) Build conserved core framework Rigid Body assembly

35. use the “spare part” algorithm to find compatible fragments in a Loop-Database “ab-initio” rebuilding of loops (Monte Carlo, molecular dynamics, genetic algorithms, etc.) b) Loop modeling Rigid Body assembly

36. EF-Hand Calcium binding aa{baalal}bbXh{DXDpDG}Xh aa{baalal}bb Xh{DXDpDG}Xh D 100% D/N 100%

37. P-loop GTP binding Conformation bb{eppgag}aa Motif sequence hh{GhXXpG}Kp P-loop GTP binding Conformation bb{eppgag}aa Motif sequence hh{GhXXpG}Kp

38. NAD(P)/FADbinding bb{eab}aahh{GhG}hX bb{eab}aa hh{GhG}hX

39. Ring closure search Distance restraints

40. Building CA Superposition

41. Building

42. A. Sali & T. Blundel (1993) JMB 234, 779 - 815 Modeling by Satisfaction of Spatial restraints M A T E A F T S G Q

43. Feature properties can be associated with a protein (e.g. X-ray resolution) residues (e.g. solvent accessibility) pairs of residues (e.g. C a - C a distance) other features (e.g. main chain classes) How can we derive modeling restraints from this data? A restraint is defined as probability density function (pdf) p(x): with Modeling by Satisfaction of Spatial restraints

44. combine basis pdfs to molecular probability density functions Modeling by Satisfaction of Spatial restraints

45. Satisfaction of spatial restraints Find the protein model with the highest probability Variable target function: Start with e.g. a random conformation model and use only local restraints minimize some steps using a conjugate gradient optimization repeat with introducing more and more long range restraints until all restraints are used Modeling by Satisfaction of Spatial restraints

46. Threading Idea: Find the optimal structure for a new (target) sequence in the set of known 3D-structures (templates) by threading the target sequence.

47. Nº sequences > Nº folds Classification - Homologs - Remotes - Analogs Step 1 Selection of templates FOLD ASSIGNMENT Structure 3D

48. Pseudo Potencial Nº sequences > Nº folds Structure 3D Classification - Homologs - Remotes - Analogs

49. Pseudo-potencial Principles Potential of Mean Force: Applied on proteins Boltzman law Use frequencies instead of probabilities. They are taken from non-homologous proteins on the PDB.

50. Fold recognition / Threading Principle: Find a compatible fold for a given sequence.... >Protein XY MSTLYEKLGGTTAVDLAV DKFYERVLQDDRIKHFFA DVDMAKQRAHQKAFLTYA FGGTDKYDGRYMREAHKE LVENHGLNGEHFDAVAED LLATLKEMGVPEDLIAEV AAVAGAPAHKRDVLNQ  ? Using... 1D – 3D profile matching, secondary structure predictions, position specific scoring matrices (PSSM), mean force potentials, keyword statistics,....

51. Predicción de “fold” Pseudo Potencial Nº sequences > Nº folds Structure 3D Classification - Homologs - Remotes - Analogs “fold” prediction

52. Predicción del “fold” Combine sequence with all folds Evaluate energies Data Base PDB Data Base PDB Energy < threshold ? Sequence Are energy profiles native-like ? YES NO

53. Fold recognition from Secondary structure prediction TOPITS

54. MODEL Predicción de “fold” Pseudo Potencial Nº sequences > Nº folds Structure 3D Classification - Homologs - Remotes - Analogs “fold” prediction

55. Hidropaticity Plot 2D prediction MODEL Predicción de “fold” Pseudo Potencial Nº sequences > Nº folds Structure 3D Classification - Homologs - Remotes - Analogs “fold” prediction 2D Structure Comparative modelling - Composer - Modeler SCR & VR SCR VR - Ring Closure - DB Search - Classification Alignments

56. Hidropaticity Plot 2D prediction Evaluate de model MODEL Predicción de “fold” Pseudo Potencial Nº sequences > Nº folds Structure 3D Classification - Homologs - Remotes - Analogs “fold” prediction 2D Structure Comparative modelling - Composer - Modeler SCR & VR SCR VR - Ring Closure - DB Search - Classification Alignments

57. Why Model Evaluation ? Just because it’s pretty doesn’t make it right ! Topics: correct fold model coverage (%) C  - deviation (rmsd) alignment accuracy (%) side chain placement

58. 1.Errors in side-chain packing. 2.Shifts of correctly aligned residues. 3.Regions without template. 4.Errors due to misalignments. 5.Errors produced by incorrect templates. Types of Errors

59. EVA Evaluation of Automatic protein structure prediction [ Burkhard Rost, Andrej Sali, http://maple.bioc.columbia.edu/eva/ ] CASP Community Wide Experiment on the Critical Assessment of Techniques for Protein Structure Prediction http://PredictionCenter.llnl.gov/casp5/ 3D - Crunch Very Large Scale Protein Modelling Project http://www.expasy.org/swissmod/SM_LikelyPrecision.html Model Accuracy Evaluation

60. Hidropaticity Plot 2D prediction Evaluate de model MODEL Predicción de “fold” Pseudo Potencial Nº sequences > Nº folds Structure 3D Classification - Homologs - Remotes - Analogs “fold” prediction 2D Structure Comparative modelling - Composer - Modeler SCR & VR SCR VR - Ring Closure - DB Search - Classification Alignments

61. Hidropaticity Plot 2D prediction Evaluate de model MODEL Predicción de “fold” Pseudo Potencial Nº sequences > Nº folds Structure 3D Classification - Homologs - Remotes - Analogs “fold” prediction 2D Structure Comparative modelling - Composer - Modeler SCR & VR SCR VR - Ring Closure - DB Search - Classification Alignments Side-chains - Multi-Copy

62. c) Side Chain placement Find the most probable side chain conformation, using homologues structures back-bone dependent rotamer libraries energetic and packing criteria Rigid Body assembly

63. Hidropaticity Plot 2D prediction Evaluate de model MODEL Predicción de “fold” Pseudo Potencial Nº sequences > Nº folds Structure 3D Classification - Homologs - Remotes - Analogs “fold” prediction 2D Structure Comparative modelling - Composer - Modeler SCR & VR SCR VR - Ring Closure - DB Search - Classification Alignments Side-chains - Multi-Copy

64. Hidropaticity Plot 2D prediction Evaluate de model MODEL Predicción de “fold” Pseudo Potencial Nº sequences > Nº folds Structure 3D Classification - Homologs - Remotes - Analogs “fold” prediction 2D Structure Comparative modelling - Composer - Modeler SCR & VR SCR VR - Ring Closure - DB Search - Classification Alignments Side-chains - Multi-Copy Optimization - Minimization E - Simulated Annealing

65. modeling will produce unfavorable contacts and bonds  idealization of local bond and angle geometry extensive energy minimization will move coordinates away  keep it to a minimum SwissModel is using GROMOS 96 force field for a steepest descent d) Energy minimization Rigid Body assembly

66. OPTIMIZATION 1- Minimization by Newton-Rapson X i+1 = X i + grad(E) 2- Simulated Annealing

67. Optimization schedule and progress [ A. Sali & T. Blundel (1993) JMB 234, 779 – 815 ] Modeling by Satisfaction of Spatial restraints

68. Hidropaticity Plot 2D prediction Evaluate de model MODEL Predicción de “fold” Pseudo Potencial Nº sequences > Nº folds Structure 3D Classification - Homologs - Remotes - Analogs “fold” prediction 2D Structure Comparative modelling - Composer - Modeler SCR & VR SCR VR - Ring Closure - DB Search - Classification Alignments Side-chains - Multi-Copy Optimization - Minimization E - Simulated Annealing

69. Hidropaticity Plot 2D prediction Evaluate de model MODEL Predicción de “fold” Pseudo Potencial Nº sequences > Nº folds Structure 3D Classification - Homologs - Remotes - Analogs “fold” prediction 2D Structure Comparative modelling - Composer - Modeler SCR & VR SCR VR - Ring Closure - DB Search - Classification Alignments Side-chains - Multi-Copy Optimization - Minimization E - Simulated Annealing Molecular Dynamics

70. MOLECULAR DYNAMICS Finite differences algorithm Coupling to a thermal bath Statistical Mechanics Ergodic Hipothesis

71. EXAMPLE

72. clevage clevage + Activation segment activation Enzime PRO-CARBOXIPEPTIDASA

73. PRO-CARBOXYPEPTIDASES Porcine Pro Carboxypeptidase B PCPBpPorcine Pro Carboxypeptidase A1 PCPA1pBovine PCPA1b

74. 10 20 30 40 50 60 70 80 PCPA2h -ETFVGDQVLEIVPSNEEQIKNLLQLEAQEHLQLDFWKSPTTPGETAHVRVPFVNVQAVKVFLESQGIAYSIMIEDVQVL PCPA2h PCPA1b KEDFVGHQVLRITAADEAEVQTVKELEDLEHLQLDFWRGPGQPGSPIDVRVPFPSLQAVKVFLEAHGIRYRIMIEDVQSL PCPA1b PCPA1p KEDFVGHQVLRISVDDEAQVQKVKELEDLEHLQLDFWRGPARPGFPIDVRVPFPSIQAVKVFLEAHGIRYTIMIEDVQLL PCPA1p 90 100 110 120 130 140 150 160 PCPA2h LDKENEEMLFNRRRERSGN-FNFGAYHTLEEISQEMDNLVAEHPGLVSKVNIGSSFENRPMNVLKFSTGG-DKPAIWLDA PCPA2h PCPA1b LDEEQEQMFASQSRARSTNTFNYATYHTLDEIYDFMDLLVAEHPQLVSKLQIGRSYEGRPIYVLKFSTGGSNRPAIWIDL PCPA1b PCPA1p LDEEQEQMFASQGRARTTSTFNYATYHTLEEIYDFMDILVAEHPALVSKLQIGRSYEGRPIYVLKFSTGGSNRPAIWIDS PCPA1p 170 180 190 200 210 220 230 240 PCPA2h GIHAREWVTQATALWTANKIVSDYGKDPSITSILDALDIFLLPVTNPDGYVFSQTKNRMWRKTRSKVSGSLCVGVDPNRN PCPA2h PCPA1b GIHSREWITQATGVWFAKKFTEDYGQDPSFTAILDSMDIFLEIVTNPDGFAFTHSQNRLWRKTRSVTSSSLCVGVDANRN PCPA1b PCPA1p GIXSRXWITQASGVWFAKKITENYGQNSSFTAILDSMDIFLEIVTNPNGFAFTHSDNRLWRKTRSKASGSLCVGSDSNRN PCPA1p 250 260 270 280 290 300 310 320 PCPA2h WDAGFGGPGASSNPCSDSYHGPSANSEVEVKSIVDFIKSHGKVKAFIILHSYSQLLMFPYGYKCTKLDDFDELSEVAQKA PCPA2h PCPA1b WDAGFGKAGASSSPCSETYHGKYANSEVEVKSIVDFVKDHGNFKAFLSIHSYSQLLLYPYGYTTQSIPDKTELNQVAKSA PCPA1b PCPA1p WDAGFGGAGASSSPCAETYHGKYPNSEVEVKSITDFVKNNGNIKAFISIXSYSQLLLYPYGYKTQSPADKSELNQIAKSA PCPA1p 330 340 350 360 370 380 390 400 PCPA2h AQSLRSLHGTKYKVGPICSVIYQASGGSIDWSYDYGIKYSFAFELRDTGRYGFLLPARQILPTAEETWLGLKAIMEHVRD PCPA2h PCPA1b VEALKSLYGTSYKYGSIITTIYQASGGSIDWSYNQGIKYSFTFELRDTGRYGFLLPASQIIPTAQETWLGVLTIMEHTLN PCPA1b PCPA1p VAALKSLYGTSYKYGSIITVIYQASGGVIDWTYNQGIKYSFSFELRDTGRRGFLLPASQIIPTAQETWLALLTIMEHTLN PCPA1p SEQUENCE ALIGNMENT OF PRO-CARBOXYPEPTIDASES

75. ComparativeModelling Model building of conserved regions Baldomero Oliva Miguel UniversitatPompeuFabra

76. UniversitatPompeuFabra ComparativeModelling Model building of conserved regions

77. Add and model build loop regions Baldomero Oliva Miguel UniversitatPompeuFabra ComparativeModelling Model building of conserved regions

78. Baldomero Oliva Miguel UniversitatPompeuFabra Add and model build loop regions ComparativeModelling Model building of conserved regions

79. Secondary Structure Prediction and Multiple Alignment of Pro-Carboxypeptidases 10 20 30 40 50 60 70 80 EEEEE HHHHHHHHHHHHHHH EEEE EEEEE HHHHHHHHHH EEEEHHHHHHHH PCPA2h PHD PCPA2h -ETFVGDQVLEIVPSNEEQIKNLLQLEAQEHLQLDFWKSPTTPGETAHVRVPFVNVQAVKVFLESQGIAYSIMIEDVQVL PCPA2h PCPA1b KEDFVGHQVLRITAADEAEVQTVKELEDLEHLQLDFWRGPGQPGSPIDVRVPFPSLQAVKVFLEAHGIRYRIMIEDVQSL PCPA1b PCPA1p KEDFVGHQVLRISVDDEAQVQKVKELEDLEHLQLDFWRGPARPGFPIDVRVPFPSIQAVKVFLEAHGIRYTIMIEDVQLL PCPA1p 90 100 110 120 130 140 150 160 HHHHHHHHHHHHHH PCPA2h PHD PCPA2h LDKENEEMLFNRRRERSGN-FNFGAYHTLEEISQEMDNLVAEHPGLVSKVNIGSSFENRPMNVLKFSTGG-DKPAIWLDA PCPA2h PCPA1b LDEEQEQMFASQSRARSTNTFNYATYHTLDEIYDFMDLLVAEHPQLVSKLQIGRSYEGRPIYVLKFSTGGSNRPAIWIDL PCPA1b PCPA1p LDEEQEQMFASQGRARTTSTFNYATYHTLEEIYDFMDILVAEHPALVSKLQIGRSYEGRPIYVLKFSTGGSNRPAIWIDS PCPA1p 170 180 190 200 210 220 230 240 PCPA2h GIHAREWVTQATALWTANKIVSDYGKDPSITSILDALDIFLLPVTNPDGYVFSQTKNRMWRKTRSKVSGSLCVGVDPNRN PCPA2h PCPA1b GIHSREWITQATGVWFAKKFTEDYGQDPSFTAILDSMDIFLEIVTNPDGFAFTHSQNRLWRKTRSVTSSSLCVGVDANRN PCPA1b PCPA1p GIXSRXWITQASGVWFAKKITENYGQNSSFTAILDSMDIFLEIVTNPNGFAFTHSDNRLWRKTRSKASGSLCVGSDSNRN PCPA1p 250 260 270 280 290 300 310 320 PCPA2h WDAGFGGPGASSNPCSDSYHGPSANSEVEVKSIVDFIKSHGKVKAFIILHSYSQLLMFPYGYKCTKLDDFDELSEVAQKA PCPA2h PCPA1b WDAGFGKAGASSSPCSETYHGKYANSEVEVKSIVDFVKDHGNFKAFLSIHSYSQLLLYPYGYTTQSIPDKTELNQVAKSA PCPA1b PCPA1p WDAGFGGAGASSSPCAETYHGKYPNSEVEVKSITDFVKNNGNIKAFISIXSYSQLLLYPYGYKTQSPADKSELNQIAKSA PCPA1p 330 340 350 360 370 380 390 400 PCPA2h AQSLRSLHGTKYKVGPICSVIYQASGGSIDWSYDYGIKYSFAFELRDTGRYGFLLPARQILPTAEETWLGLKAIMEHVRD PCPA2h PCPA1b VEALKSLYGTSYKYGSIITTIYQASGGSIDWSYNQGIKYSFTFELRDTGRYGFLLPASQIIPTAQETWLGVLTIMEHTLN PCPA1b PCPA1p VAALKSLYGTSYKYGSIITVIYQASGGVIDWTYNQGIKYSFSFELRDTGRRGFLLPASQIIPTAQETWLALLTIMEHTLN PCPA1p

80. Pseudo-energy(threading) Wrong model Evaluation

81. G (C’) N (C’’) S (Ccap ) R (C1) E (C2) R (C3) R (C5) R (C4) Schellman Motif Profile C´´ » C3 / C´ Gly C3 C2 C1 Ccap C’ C’’ L Q E H G I L Q E H G I A R A K G V A R A K G V M K K L G K M K K L G K pont d’ hidrògen Interacció hidrofòbica N R R R E R S G N F N ponts d’hidrògen C3 Ccap C’’ ^ ^ ^  -Helix C-cap Schellman Motif Schellman Motif

82. Refinement of the Model

83. pseudo-energy residue IMPROVED

84. Protein Structure Resources PDBhttp://www.pdb.org PDB – Protein Data Bank of experimentally solved structures (RCSB) CATHhttp://www.biochem.ucl.ac.uk/bsm/cath/ Hierarchical classification of protein domain structures SCOPhttp://scop.mrc-lmb.cam.ac.uk/scop/ Alexey Murzin’s Structural Classification of proteins DALIhttp://www2.ebi.ac.uk/dali/ Lisa Holm and Chris Sander’s protein structure comparison server SS-Prediction and Fold Recognition PHDhttp://cubic.bioc.columbia.edu/predictprotein/ Burkhard Rost’s Secondary Structure and Solvent Accessibility Prediction Server 3DPSSMhttp://www.sbg.bio.ic.ac.uk/~3dpssm/ Fold Recognition Server using 1D and 3D Sequence Profiles coupled with Secondary Structure and Solvation Potential Information.

85. Protein Homology Modeling Resources SWISS MODEL: http://www.expasy.ch/swissmod/ Deep View - SPDBV: homepage: http://www.expasy.ch/spdbv/ Tutorials http://www.usm.maine.edu/~rhodes/SPVTut/ http://www.bbsrc.ac.uk/molbiol/ WhatIfhttp://www.cmbi.kun.nl/whatif/ Gert Vriend’s protein structure modeling analysis program WhatIf Modeller: http://guitar.rockefeller.edu/modeller/ Andrej Sali's homology protein structure modelling by satisfaction of spatial restraints FAMS: http://physchem.pharm.kitasato-u.ac.jp/FAMS/fams.html Full Automatic Modelling System (FAMS); Kitasato University; Tokyo, Japan 3D-JIGSAW: http://www.bmm.icnet.uk/people/paulb/3dj/form.html Comparative Modelling Server; Imperial Cancer Research Fund; London, UK CPHmodels: http://www.cbs.dtu.dk/services/CPHmodels/ Centre for Biological Sequence Analysis; The Technical University of Denmark; Denmark SDSC1: http://cl.sdsc.edu/hm.html SDSC Structure Homology Modelling Server; San Diego Supercomputing Centre

86. Protein Homology Modeling Resources SWISS MODEL: http://www.expasy.ch/swissmod/ Example Dear baldo, Title of your Request: subtilisin Swiss-Model '(SWISS-MODEL Version 36.0003)' started on 27-Oct-2003 Process identification is AAAa08uiF The modelling procedure is now in progress, and its results should be sent to you shortly. In case of difficulties, please contact: Torsten Schwede Nicolas Guex

87. Protein Homology Modeling Resources SWISS MODEL: http://www.expasy.ch/swissmod/ Example Length of target sequence: 379 residues Searching sequences of known 3D structures Found 1c3lA.pdb with P(N)=0 Found 1be8_.pdb with P(N)=0 Found 1cseE.pdb with P(N)=0 Found 1scd_.pdb with P(N)=0 Found 1sbc_.pdb with P(N)=0 Found 1be6_.pdb with P(N)=0 …….

88. Protein Homology Modeling Resources 3D JigSaw Example This is an automatic split of your sequence in protein domains Follow the link below and select the modelling templates for each modelable domain 2 domain/s found in your query sequence (2 have homologous templates). See them at: http://www.bmm.icnet.uk/~3djigsaw/dom_fish/display2.cgi?output/590e3988 (this file will be erased in a week time) Your submission: >subtilisin MMRKKSFWFGMLTAFMLVFTMEFSDSASAAQPGKNVEKDYFVGFKSGVKTASVKKDIIKE SGGKVDKQFRIINAAKATLDKEALKEVKNDPDVAYVEEDHVAHALGQTVPYGIPLIKADK VQAQGFKGANVKVAVLDTGIQASHPDLNVVGGASFVAGEAYNTDGNGHGTHVAGTVAALD NTTGVLGVAPSVSLFAVKVLNSSGSGSYSGIVSGIEWATTNGMDVINMSLGGPSGSTAMK QAVDNAYSKGVVPVAAAGNSGSSGYTNTIGYPAKYDSVIAVGAVDSNSNRASFSSVGAEL EVMAPGAGVYSTYPTNTYATLNGTSMASPHVAGAAALILSKHPNLSASQVRNRLSSTATY LGSSFYYGKGLINVEAAAQ Send comments mailto:3djigsaw@cancer.org.uk See you soon http://www.bmm.icnet.uk/~3djigsaw/