Introduction to Bioinformatics 20120

1
Introduction to Bioinformatics20120

• Gianluca Pollastri
• office CS A1.07
• email gianluca.pollastri_at_ucd.ie

2
Credits

• Richard Lathrop and Pierre Baldis Bioinformatics
  courses at University of California _at_ Irvine.

3
Credits (2)

• PSI-BLAST tutorial on the NCBI web site

4
Course overview

• Context DNA, RNA, proteins
• Resources GenBank, PDB, etc.
• Algorithms for sequence comparison.
• Phylogenetics.
• Structural bioinformatics protein structure
  prediction.

5
Lecture notes

• http//gruyere.ucd.ie/2007_courses/20120/
• confidential..

6
Recommended/useful readings

• No book is actually required
• Introduction to Bioinformatics
• Lesk
• Introduction to Computational Molecular Biology
• Setubal, Meidanis
• Bioinformatics the Machine Learning approach
• Baldi, Brunak

7
BLAST

• Basic Local Alignment Search Tool
• Popular package for sequence vs whole database of
  sequences comparisons.
• Approximates Smith-Waterman (local alignment)
• But while SW would take ages in many practical
  cases (sequence_length2 x _sequences), BLAST is
  very fast (sequence_length x _sequences)

8
PSI-BLAST

• Position-Specific Iterated BLAST.
• The substitution matrix is not fixed anymore. A
  position specific scoring matrix (PSSM) is
  constructed automatically from a multiple
  alignment of the highest scoring hits in an
  initial BLAST search
• This process can be iterated any number of times

9
PSI-BLAST (2)

• The PSSM is generated by calculating
  position-specific scores for each position in the
  alignment.
• Highly conserved positions receive high scores
  and weakly conserved positions receive scores
  near zero (similar to building standard matrix,
  except that here the score depends on the
  position).
• The PSSM is used to perform a second (etc.) BLAST
  search and the results of each iteration used to
  refine the PSSM.
• This iterative searching strategy results in
  increased sensitivity, because the
  penalties/rewards for substitutions are specific
  to a set, or family of proteins.

10

• Last position-specific scoring matrix computed,
  weighted observed percentages rounded down,
  information per position, and relative weight of
  gapless real matches to pseudocounts
• A R N D C Q E G H I L K M
  F P S T W Y V A R N D C Q E
  G H I L K M F P S T W Y V
• 1 M -1 -2 -3 -4 -2 -1 -3 -3 -2 1 1 -2 7
  -1 -3 -2 -1 -2 -2 2 0 0 0 0 0 0 0
  0 0 0 0 0 77 0 0 0 0 0 0
  23 0.62 0.18
• 2 N 0 -1 3 -1 -2 -1 -1 -1 -1 -2 -3 -1 -2
  -3 -2 3 4 -4 -2 -2 0 0 21 0 0 0 0
  0 0 0 0 0 0 0 0 40 39 0 0
  0 0.55 0.22
• 3 I -2 -4 -4 -4 -2 -3 -4 -5 -4 5 2 -3 3
  -1 -4 -3 -2 -3 -2 2 0 0 0 0 0 0 0
  0 0 75 14 0 11 0 0 0 0 0 0
  0 0.71 0.28
• 4 F 0 -1 -1 0 -3 0 4 -3 -1 -3 -3 2 -2
  2 -2 -1 1 -3 -1 -2 11 0 0 0 0 0 49
  0 0 0 0 15 0 15 0 0 10 0 0
  0 0.46 0.28
• 5 E -2 -1 0 5 -4 2 4 -3 -1 -4 -4 2 -3
  -4 -2 -1 -2 -4 -3 -4 0 0 0 41 0 10 30
  0 0 0 0 19 0 0 0 0 0 0 0
  0 0.78 0.32
• 6 M -2 -2 -3 -4 -2 -1 -3 -4 -3 0 1 -2 9
  -1 -4 -2 -2 -2 -2 0 0 0 0 0 0 0 0
  0 0 0 0 0 100 0 0 0 0 0 0
  0 1.14 0.33
• 7 L -2 -3 -4 -5 -2 -3 -4 -5 -4 1 5 -4 1
  0 -4 -3 -2 -3 -2 0 0 0 0 0 0 0 0
  0 0 0 100 0 0 0 0 0 0 0 0
  0 0.92 0.33
• 8 R -2 5 -2 -3 -4 0 -1 -3 -2 -2 -2 3 -2
  -3 -3 -2 -2 -4 -3 1 0 53 0 0 0 0 0
  0 0 0 0 28 0 0 0 0 0 0 0
  19 0.71 0.33
• 9 I -2 2 1 -2 -4 3 -1 0 0 0 -2 -1 -2
  0 -3 -2 -2 -1 6 -2 0 16 9 0 0 18 0
  9 0 10 0 0 0 0 0 0 0 0 38
  0 0.48 0.33
• 10 D -3 -3 0 7 -5 -1 1 -2 -2 -4 -5 -2 -4
  -5 -2 -1 -2 -5 -4 -4 0 0 0 100 0 0 0
  0 0 0 0 0 0 0 0 0 0 0 0
  0 1.47 0.33
• 11 E -2 -1 -1 1 -5 1 6 -3 -1 -4 -4 0 -3
  -4 -2 -1 -2 -4 -3 -3 0 0 0 0 0 0 100
  0 0 0 0 0 0 0 0 0 0 0 0
  0 1.22 0.33
• 12 G -1 -3 -1 -2 -4 -3 -3 7 -3 -5 -5 -3 -4
  -4 -3 -1 -3 -4 -4 -4 0 0 0 0 0 0 0
  100 0 0 0 0 0 0 0 0 0 0 0
  0 1.51 0.33
• 13 L -2 1 -3 -2 -3 -1 2 -4 -2 1 3 -2 0
  0 -3 -2 -2 -2 4 1 0 10 0 0 0 0 18
  0 0 9 35 0 0 0 0 0 0 0 18
  9 0.33 0.33
• 14 R -2 6 -1 -2 -4 0 -1 -3 -1 -4 -3 4 -2
  -4 -3 -1 -2 -4 -3 -3 0 62 0 0 0 0 0
  0 0 0 0 38 0 0 0 0 0 0 0
  0 1.00 0.33
• 15 L -2 -3 -3 1 -3 -2 2 -4 -3 0 4 -2 0
  -1 -3 -2 0 -3 -2 1 0 0 0 8 0 0 18
  0 0 0 56 0 0 0 0 0 9 0 0
  9 0.39 0.33
• 16 K -1 0 -1 -1 -3 0 2 -3 -2 -2 -3 5 -2
  -4 -2 1 1 -4 -3 0 0 0 0 0 0 0 18
  0 0 0 0 54 0 0 0 9 9 0 0
  9 0.52 0.33
• 17 I -2 -3 -4 -4 -2 -3 -4 -4 -4 2 1 -3 4
  -1 -4 -3 -2 7 -1 4 0 0 0 0 0 0 0
  0 0 10 9 0 18 0 0 0 0 18 0
  45 0.71 0.33

11
PSI-BLAST (3)

• Most parameters are similar to BLAST
• An important new option is -k. This is the
  maximum e-value (expectation that a match
  occurred by chance) for a sequence to be included
  into the PSSM.
• Normally -k is set to stricter values (often
  10-10 or smaller) than -e, to ensure that the
  proteins that determine the substitution model
  for the next round are truely close to the query.

12
PSI-BLAST (4)

• Summarising
• Search for sequences that are similar to the
  query, align them
• Compute the substitution model for each position,
  based on the alignment above
• if satisfied, stop, otherwise go back to 1
• 3-4 rounds is generally considered to be ok. More
  can introduce problems.

13
PSI-BLAST (5)

• PSI-BLAST can find hits at similarity levels that
  BLAST would consider too low to be reliable
  remote homologues.
• (two proteins are homologous if they are
  evolutionarily related - in which case they are
  often structurally and functionally related)
• Hits at 10-15 sequence similarity can sometimes
  be found by PSI-BLAST.
• Still a long way to go to find all homology the
  average sequence similarity between structural
  homologues is 8-10, barely above pure chance..

14
End of pairwise sequence alignments...

• yippie!

15
Multiple sequence alignments (MA)

• We may want to find the optimal alignment of
  multiple sequences instead of pairs of sequences.
• For instance, we have proteins with the same
  function for multiple organisms we want to find
  out which parts of the sequences match and which
  parts contain most gaps and mismatches.

16
MA (2)

• Most informative alignment will contain a mixture
  of closely related and distantly related
  sequences
• if all closely related, not much information, not
  much to learn (we are looking at the result of a
  short evolutionary stretch)
• if all very remotely related, hard to even get an
  alignment (unless structures, or other bits of
  information are available)

17
What do we use MA for?

• E.g. compress them into profiles, i.e. tables of
  frequencies of various residues in different
  positions across a family of proteins.
• Profiles can be used to
• retrieve remote homologues (as in PSI-BLAST)
• identify active sites (highly conserved)
• identify surface loops (can help for instance to
  design vaccines)
• predict more accurately just about anything about
  a protein more information in MA (evolutionary
  snapshot) than in single sequence

18

• Last position-specific scoring matrix computed,
  weighted observed percentages rounded down,
  information per position, and relative weight of
  gapless real matches to pseudocounts
• A R N D C Q E G H I L K M
  F P S T W Y V A R N D C Q E
  G H I L K M F P S T W Y V
• 1 M -1 -2 -3 -4 -2 -1 -3 -3 -2 1 1 -2 7
  -1 -3 -2 -1 -2 -2 2 0 0 0 0 0 0 0
  0 0 0 0 0 77 0 0 0 0 0 0
  23 0.62 0.18
• 2 N 0 -1 3 -1 -2 -1 -1 -1 -1 -2 -3 -1 -2
  -3 -2 3 4 -4 -2 -2 0 0 21 0 0 0 0
  0 0 0 0 0 0 0 0 40 39 0 0
  0 0.55 0.22
• 3 I -2 -4 -4 -4 -2 -3 -4 -5 -4 5 2 -3 3
  -1 -4 -3 -2 -3 -2 2 0 0 0 0 0 0 0
  0 0 75 14 0 11 0 0 0 0 0 0
  0 0.71 0.28
• 4 F 0 -1 -1 0 -3 0 4 -3 -1 -3 -3 2 -2
  2 -2 -1 1 -3 -1 -2 11 0 0 0 0 0 49
  0 0 0 0 15 0 15 0 0 10 0 0
  0 0.46 0.28
• 5 E -2 -1 0 5 -4 2 4 -3 -1 -4 -4 2 -3
  -4 -2 -1 -2 -4 -3 -4 0 0 0 41 0 10 30
  0 0 0 0 19 0 0 0 0 0 0 0
  0 0.78 0.32
• 6 M -2 -2 -3 -4 -2 -1 -3 -4 -3 0 1 -2 9
  -1 -4 -2 -2 -2 -2 0 0 0 0 0 0 0 0
  0 0 0 0 0 100 0 0 0 0 0 0
  0 1.14 0.33
• 7 L -2 -3 -4 -5 -2 -3 -4 -5 -4 1 5 -4 1
  0 -4 -3 -2 -3 -2 0 0 0 0 0 0 0 0
  0 0 0 100 0 0 0 0 0 0 0 0
  0 0.92 0.33
• 8 R -2 5 -2 -3 -4 0 -1 -3 -2 -2 -2 3 -2
  -3 -3 -2 -2 -4 -3 1 0 53 0 0 0 0 0
  0 0 0 0 28 0 0 0 0 0 0 0
  19 0.71 0.33
• 9 I -2 2 1 -2 -4 3 -1 0 0 0 -2 -1 -2
  0 -3 -2 -2 -1 6 -2 0 16 9 0 0 18 0
  9 0 10 0 0 0 0 0 0 0 0 38
  0 0.48 0.33
• 10 D -3 -3 0 7 -5 -1 1 -2 -2 -4 -5 -2 -4
  -5 -2 -1 -2 -5 -4 -4 0 0 0 100 0 0 0
  0 0 0 0 0 0 0 0 0 0 0 0
  0 1.47 0.33
• 11 E -2 -1 -1 1 -5 1 6 -3 -1 -4 -4 0 -3
  -4 -2 -1 -2 -4 -3 -3 0 0 0 0 0 0 100
  0 0 0 0 0 0 0 0 0 0 0 0
  0 1.22 0.33
• 12 G -1 -3 -1 -2 -4 -3 -3 7 -3 -5 -5 -3 -4
  -4 -3 -1 -3 -4 -4 -4 0 0 0 0 0 0 0
  100 0 0 0 0 0 0 0 0 0 0 0
  0 1.51 0.33
• 13 L -2 1 -3 -2 -3 -1 2 -4 -2 1 3 -2 0
  0 -3 -2 -2 -2 4 1 0 10 0 0 0 0 18
  0 0 9 35 0 0 0 0 0 0 0 18
  9 0.33 0.33
• 14 R -2 6 -1 -2 -4 0 -1 -3 -1 -4 -3 4 -2
  -4 -3 -1 -2 -4 -3 -3 0 62 0 0 0 0 0
  0 0 0 0 38 0 0 0 0 0 0 0
  0 1.00 0.33
• 15 L -2 -3 -3 1 -3 -2 2 -4 -3 0 4 -2 0
  -1 -3 -2 0 -3 -2 1 0 0 0 8 0 0 18
  0 0 0 56 0 0 0 0 0 9 0 0
  9 0.39 0.33
• 16 K -1 0 -1 -1 -3 0 2 -3 -2 -2 -3 5 -2
  -4 -2 1 1 -4 -3 0 0 0 0 0 0 0 18
  0 0 0 0 54 0 0 0 9 9 0 0
  9 0.52 0.33
• 17 I -2 -3 -4 -4 -2 -3 -4 -4 -4 2 1 -3 4
  -1 -4 -3 -2 7 -1 4 0 0 0 0 0 0 0
  0 0 10 9 0 18 0 0 0 0 18 0
  45 0.71 0.33

19
Rules about MA

• Inserting spaces in sequences so that all their
  lengths are the same. Illegal (pointless) all
  sequences have a space in the same position.
• Find the MA that has the best score according to
  some criterion.

20
SP measure

• Scoring alignments isnt as trivial as in the
  pairwise case.
• SP (sum-of-pairs) is a reasonable scoring
  function
• where k are the positions of the alignment, and
  s_i and s_j are sequences i and j, as aligned
  (with gaps).

21
Example

• Alignment of Coronavirus sequences and 1 SARS
  sequence
• Murine_hepatitis_vir_AF207902
  CCATACCGGCGTATGCGAAGCAGTGGTT-GCAACCCTGGTCCATCCTTCT
  
• Bovine_coronavirus___NC_003045
  TTATGCCTGTGCAATCCCGGAAATTTAT-TGTTCCTTGGGTTATGTACTT
  
• Avian_infectious_bro_NC_001451
  CTAGCCTTGCGCTAGATTTTTAACTTA--ACAAAACGGACTTAAATACCT
  
• Porcine_epidemic_dia_NC_003436
  TAGGTTTGCTTAAGTAGCCATCGCAAGT-GCTGTGCTGTCCTCTAGTTCC
  
• Human_coronavirus_22_NC_002645
  TTGGGTTGCAACAGTT-TGGAAGCAAGT-GCTGTG-TGTCCTAGTCTAAG
  
• Transmissible_gastro_NC_002306
  TCAGTTTG--GCAATC--ACTCCTTGGA-ACGGGGTTGAGCGAACGGTGC
  
• gi30468046gbAY283798.1
  TAAACGTTCTGATGCCTTAAGCACCAATCACGGCCACAAGGTCGTTGAGC
  
• First column
• p(C,T) p(C,C) p(C,T) p(C,T) p(C,T)
  p(C,T)
• p(T,C) p(T,T) p(T,T) p(T,T) p(T,T)
• p(C,T) p(C,T) p(C,T) p(C,T)
• p(T,T) p(T,T) p(T,T)
• p(T,T) p(T,T)
• p(T,T)

22
p()

• We can use the same penalty function used for
  pairwise aligments
• match 1
• mismatch -1
• gap -2

23
Example

• Alignment of Coronavirus sequences and 1 SARS
  sequence
• Murine_hepatitis_vir_AF207902
  CCATACCGGCGTATGCGAAGCAGTGGTT-GCAACCCTGGTCCATCCTTCT
  
• Bovine_coronavirus___NC_003045
  TTATGCCTGTGCAATCCCGGAAATTTAT-TGTTCCTTGGGTTATGTACTT
  
• Avian_infectious_bro_NC_001451
  CTAGCCTTGCGCTAGATTTTTAACTTA--ACAAAACGGACTTAAATACCT
  
• Porcine_epidemic_dia_NC_003436
  TAGGTTTGCTTAAGTAGCCATCGCAAGT-GCTGTGCTGTCCTCTAGTTCC
  
• Human_coronavirus_22_NC_002645
  TTGGGTTGCAACAGTT-TGGAAGCAAGT-GCTGTG-TGTCCTAGTCTAAG
  
• Transmissible_gastro_NC_002306
  TCAGTTTG--GCAATC--ACTCCTTGGA-ACGGGGTTGAGCGAACGGTGC
  
• gi30468046gbAY283798.1
  TAAACGTTCTGATGCCTTAAGCACCAATCACGGCCACAAGGTCGTTGAGC
  
• First column
• (-1) 1 (-1) (-1) (-1) (-1)
• (-1) 1 1 1 1
• (-1) (-1) (-1) (-1)
• 1 1 1
• 1 1
• 1 1

24
Example

• Alignment of Coronavirus sequences and 1 SARS
  sequence
• Murine_hepatitis_vir_AF207902
  CCATACCGGCGTATGCGAAGCAGTGGTT-GCAACCCTGGTCCATCCTTCT
  
• Bovine_coronavirus___NC_003045
  TTATGCCTGTGCAATCCCGGAAATTTAT-TGTTCCTTGGGTTATGTACTT
  
• Avian_infectious_bro_NC_001451
  CTAGCCTTGCGCTAGATTTTTAACTTA--ACAAAACGGACTTAAATACCT
  
• Porcine_epidemic_dia_NC_003436
  TAGGTTTGCTTAAGTAGCCATCGCAAGT-GCTGTGCTGTCCTCTAGTTCC
  
• Human_coronavirus_22_NC_002645
  TTGGGTTGCAACAGTT-TGGAAGCAAGT-GCTGTG-TGTCCTAGTCTAAG
  
• Transmissible_gastro_NC_002306
  TCAGTTTG--GCAATC--ACTCCTTGGA-ACGGGGTTGAGCGAACGGTGC
  
• gi30468046gbAY283798.1
  TAAACGTTCTGATGCCTTAAGCACCAATCACGGCCACAAGGTCGTTGAGC
  
• What about columns with multiple gaps?
• What is p(-,-)?
• Usually 0..

25
Property of SP

• If the gap-gap penalty is set to zero, then the
  SP score of an MA is equal to the sum of all
  pairwise scores induced by it.
• We can either sum over all pairs for each column,
  or for each pair of sequences (throwing away
  simultaneous gaps) for all pairs. This just means
  inverting the sums in

26
MA

• How can we obtain the actual MA?
• Dynamic programming, as in the pairwise case,
  would work.
• The complexity is a bit higher though..