Project list

1
Project list

• Peptide MHC binding predictions using position
  specific scoring matrices including pseudo counts
  and sequences weighting techniques
• Peptide MHC binding predictions using artificial
  neural networks with different sequence encoding
  schemes
• Gibbs sampler approach to the prediction of MHC
  class II binding motifs
• Improved protein sequence alignment using
  sequence profiles
• Improved sequence alignments using position
  specific gap penalties
• Improved protein template identification using
  hidden Markov models (HMMER)
• Chemo-informatics using Support Vector Machines
• Model of HIV infection using Cellular Automata

2
What is a Project

• Purpose
• Use a method introduced in the course to describe
  some biological problem
• How
• Construct a data set describing the problem
• Define which method to use
• (Develop method)
• Train and evaluate method
• (Compare performance to other methods)
• Documentation
• Write report in form of a research article (10-15
  pages)
• Abstract
• Introduction
• Materials and method
• Results
• Discussion
• References

3
PSSM

• Peptide MHC binding predictions using position
  specific scoring matrices including pseudo counts
  and sequences weighting techniques
• Compare methods for sequence weighting
• Clustering vs heuristics
• Benchmark (Peters et al 2006) covering some 20
  MHC molecules, compare to best other methods

4
NN

• Peptide MHC binding predictions using artificial
  neural networks with different sequence encoding
  schemes
• Benchmark (Peters et al 2006) covering some 20
  MHC molecules, compare to best other methods
• Compare sequence encoding schemes
• Sparse, Blosum, composition, charge, amino acids
  size,..

5
Gibbs sampler

• Gibbs sampler approach to the prediction of MHC
  class II binding motifs
• Develop Gibbs sampler to prediction of MHC class
  II binding motifs
• Benchmark Nielsen et al 2007 covering 14 HLA-DR
  alleles

6
Sequence alignment - 1

• Improved protein sequence alignment using
  sequence profiles
• Develop sequence profile alignment scoring scheme
  to improve sequence alignment

7
Sequence alignment - 1

• Improved protein sequence alignment using
  sequence profiles
• Benchmark (Andrej Sali. Protein Science (2004)
• SEQ1 SEQ2 ID RMSD
• 1QFE.A 1RLC.L 4.38 3.79
• 1BAR.A 1XYF.A 4.84 1.84
• 1F6Y.A 1REQ.A 6.20 3.81
• 1DTY.A 1B9H.A 8.17 3.38
• 1DTY.A 1B9H.A
• BL 1DTY.A 429 1B9H.A 388 368 84 16.6155 2
• PROF 1DTY.A 429 1B9H.A 388 429 47 6.5659 173
• Use f4 n4/qlen to evaluate methods

Qlen
Dlen
Alen
Nid
Rmsd
n4
8
Sequence alignment - 1

• Improved protein sequence alignment using
  sequence profiles
• Code example
• prf_align 2PUE.A 1BYK.A xC
• ALN 2PUE.A 340 0 1BYK.A 255 0 type prol_list alen
  269 249 -99 57 20 score 168 2246 -168 -99.900002
  rmsd -99.9 -99 -99 -99 -99 -99.9
• QAL 2PUE.A 58 KSIGLLATSSEAAYFAEIIEAVEKNCFQKGYTL
  ILGNAWNNLEKQRAYLSMMAQKRVDGLLVM-CSEYPEPLLAM
• LEEYRHIPMVVMDWGEAKADFTDAVIDNAFEGGY-MAGRYLIERGHREIG
  VIPGPLEQ-NTGAGRLAGFMKAMEEAMIKVPESWIVQGDFEP
• ESGYRAMQQILSQPHRPTAVFCGGDIMAMGALCAADEMGLRVPQDVSLIG
  YDNVRNARYFTPALTTIHQPKDSLGETAFNMLLDRIVNKREE
• PQSIEVHPRL
• DAL 1BYK.A 2 KVVAIIVTRLDSLSENLAVQTMLPAFYEQGYDP
  IMMESQFSPQLVAEHLGVLKRRNIDGVVLFGFTGITEEMLAH
• WQSS-----LVLLARDAKG-FASVCYDD--EGAIKILMQRLYDQGHRNIS
  YLGVPHSDVTTGKRRHEAYL-AFCKAH-KLHPVAALPG-LAM
• KQGYENVAKVIT-PET-TALLCATDTLALGASKYLQEQRIDTLQLAS-VG
  --NTPLMKFLHPEIVTVDPGYAEAGRQAACQLIAQVTG-RSE
• PQQIIIPATL
• prf_align 2PUE.A 1BYK.A xC aln2rmsd -- xC
  grep ALN
• ALN 2PUE.A 340 0 1BYK.A 255 0 type prol_list alen
  269 249 249 57 20 score 168 2246 -168 -99.900002
  rmsd 3.21852 113 101 173 198 0.203965

9
Sequence alignment - 2

• Improved protein sequence alignment using using
  position specific gap penalties
• Develop scheme for position specific gap
  penalties to improve sequence alignment
• Gap penalties in the core of a protein should be
  higher than gap penalties in loops

10
Hidden Markov models

• Improved protein template identification using
  hidden Markov models (HMMER)
• Train profile HMM to remote protein fold
  recognition
• Use the Hmmer program to construct profile HMM
  for selected set of proteins from the CASP8
  competition
• Use Hmmer model to identify PDB templates for
  hmology modeling for CASP8 targets

11
Support vector machines

• Chemo-informatics using Support Vector Machines
• Peptide binding using SVMs
• Compare to PSSM/NN using benchmark (Peters et al
  2006) covering some 20 MHC molecules

12
Cellular Automata

• Develop model for HIV infection using cellular
  automata