Similarity Searches on Sequence Databases

1
Similarity Searches on Sequence Databases

• Chapter 7 Page215

2
A story

• H. pylori was discover in 1984
• its genome was first sequenced in 1990s
• this was published in NATURE.
• In this publication, all proteins translated by
  the genome were also published
• HOW did they do in a short time?

3
HOW?

• They compare the sequence of the genome of H.
  pylori with those of other bacteria.
• Then they predicted the proteins of H. pylori and
  its metabolits.

4
What does this similarity mean?

• if two protein or gene sequences are similar,
  they are homologues.
• SO
• They are from similar organisms
• similar proteins means
• similar functions
• similar structures
• that is, similar charactersitics

5
How similar is very similar

• For proteins
• if gt25 identity between 2 proteins, they are
  similar

The range of identity lt25 is called the TWILIGHT
ZONE. Nothing is sure about similarity. For
nucleotides, the limit is 70
similarity (homologous)
6
Homology

• Addition to , some other information is
  essential to say that there is a homology between
  2 ones
• Expectation value less value, more homology,
• Lenght of the similar segments
• Patterns of a.a conservation
• Number of insertions/deletions

7
BLAST (Basic Local Assightment and Search Tool)

• 30 years ago, to scan the simility between our
  query and hundreds of others we would need
  several hours -(print, put on the wall, compare
  one by one manualy-)
• NOW, by speedy computers, we compare ours with
  millons at most in several minutes.

8
BLASTing Protein Sequence

• 2 strategies
• Compare
• a protein with a protein database BLASTP
• a protein with a nucleotide database TBLASTN
• (machine turns your nucleotide seq. into 6
  possible sequence)
• Important BLAST servers
• BLAST server from NCBI from USA
• BLAST server from Swiss EMBnet
• if U learn one, U use other(s)

9
(No Transcript)
10
(No Transcript)
11
Which we should choose

• Dependin on
• Database Choose the one using a database you
  want
• Speed Choose the one which is not crowded (in
  Turkey, no problem during day until 5 because US
  and Japan in dark)
• different BLAST servers return different results
  instead of the same query because of differences
  between their databases

12
BLAST output contains

• A graphic display
• A hit list
• The alighments
• The parameters

13
A graphic display

• which part of other sequences is similar to yours
• This part can be different or absent in some
  servers.
• What colors say best, good, moderate, worse,
  worst
• what does length say the same length...homologous
  , shorter corresponds to the domain

14
(No Transcript)
15
A hit list

• Accesion number (spSWISS-PROT) name
• Description You estimate whether it is
  interested or not
• Score if lt50, unreliable
• E-value lower E, more similarity
  Egt0.001.twilight zone.
• E approaching 0 is the best

16
Alignments

• Alignments say smthng on similarities btw seq
• identity gt25 is good
• lengthlength of alignment. short alignments
  gives generally high E values
• Top is ours bottom is hit () shows similar aa
• XXXXXX low complexity region
• numbers shows the coordinates

17
(No Transcript)
18
BLASTing DNA sequences

• If it is reading frame, tranlate it to protein
  than blast.
• if not choose one of them below
• a DNA from DNA BLASTN
• a TDNA from TDNA TBLASTX
• a TDNA from protein BLASTX
• Ttranslated it means blast tanslates our
  sequence into 6 possible protein sequence

19
Strategies for right choice of BLAST type for DNA
20
(No Transcript)
21
controlling blast right parameters
22
Control sequence masking

• Protein Remove low-complexity regions
• DNA many repeats. filterhuman repeats

23
BLAST output

• a less homologous sequence can be important
• WHAT? Adjust parameters
• suitable database decrease results, use swiss p.
• use the magic tags of enrez query
• Adjust E-value

24
PSI-BLAST (Position Specific Iterated-BLAST)

• BLAST finds close relatives.
• To find far relatives, use PSI-BLAST
• It uses more complex scoring procedures.