Similarity: a guide

1
Similarity a guide
2
DNA
A C G T
Four nucleotides
Any one nucleotide has a 25 chance of matching
at a given position. Therefore two random
sequences of nucleic acid will have a 25
nucleotide identity.
3
A good match
When you search a database of size X bases, what
are the chances of getting a good match ? a
sequence of 4 bases is present every 44 bases
256 bases a sequence of 6 bases is present every
64 bases 4000 bases a perfect match of 16 bases
would be expected by chance in a cDNA
library. The bigger the database, the better the
chance of getting a long match purely by chance.
4
How good is a good match ?

• Data base search engines look at the size of the
  database, the size and identity of the match,
  and calculate the likelihood of such a match
  occurring by chance.
• 1 expected to occur by random chance
• 0.01 is my minimum cutoff (10-2)
• DNA and protein sequence is not truly random the
  statistics are skewed
• Different databases calculate P differently.

5
Protein

• 20 amino acids
• Any one amino acid has a 1 in 20 chance of
  matching randomly
• Two random protein sequences will have 5
  sequence identity
• Searching a database with a protein sequence
  gives many fewer background hits, and is much
  more sensitive on statistical grounds

6
DNA or protein searches ?
DNA
Protein
Sensitivity
Poor
Much better
Size of database
Excellent
Poor
7
Significance of match
STHREBEERPMSILAGERPWYTCHERRYBRANDYSRED
QREBNNBEERALWHQLAGERNGCNRCHERRYBRANDYQQE
8
Significance of match
STHREBEERPMSILAGERPWYTCHERRYBRANDYSRED
QREBNNBEERALWHQLAGERNGCNRCHERRYBRANDYQQE
STHREBEERPMSILAGERPWYTCHERRYBRANDYSRED
QREBNNBEERALWHQLAGERNGCNRCHERRYBRANDYQQE
STHREBEERPMSILAGERPWYTCHERRYBRANDYSRED
QREBNNBEERALWHQLAGERNGCNRCHERRYBRANDYQQE
9
Similarity
DNA
Transitions
A G
C T
Transversions
C G
A T
Transitions and transversions occur at different
rates
10
Similarity- proteins
Dayhoff Matrix
Based on proteins with gt 85 identity
11
Similarity- proteins

• BLOSUM matrix
• based on proteins with low similarity
• much more sensitive for detecting related
  proteins
• Matrices based on structural models
• yet to be well described

12
Multiple Sequence Alignment

• Analysis of related proteins can demonstrate
  conserved residues, structural features of the
  protein, and identify novel facets of protein
  biology
• Prerequisite for phylogenetic analysis of
  proteins and DNA

13
MSA

• When aligning two sequences of 1000 aa, there are
  (103)2 106 possibilities
• When aligning three sequences, 109
• For nine sequences, there are 1027 possible
  alignments
• Multiple sequence alignments rely on cheats to
  make the alignment
• these alignments are not optimal
• Check manually !

14
MSA

• Align- Use ClustalX
• Check your alignment manually!
• Visualise your alignment with Genedoc or Boxshade
• allows the detection of conserved residues
• Phylogeny- go on Paul Sharps course !