Introduction to Bioinformatics

1
Introduction to Bioinformatics

• BLAST

2
BLAST

• Introduction
• What is BLAST?
• Query Sequence Formats
• What does BLAST tell you?
• Choices
• Variety of BLAST
• BLAST Programs Which One to Use?
• Commonly Used BLAST programs
• BLAST Databases Which One to Search?
• Understanding the Output
• Database Search with BLAST
• Blast Steps How It Works

Acknowledgement The presentation includes
adaptations from NCBIs Introduction to Molecular
Biology Information Resources Modules
3
What is BLAST?

• Basic Local Alignment Search Tool
• The GoogleTM of bioinformatics
• Query is a DNA or protein sequence, not a text
  term
• Character string comparison against all the
  sequences in the target database
• Rigorous statistics used to identify
  statistically significant matches

4
Query Sequence Formats

• Bare sequence
• QIKDLLVSSSTDLDTTLVLVNAIYFKGMWKTAFNAEDTREMPFHVTKQES
  KPVQMMCMNNSFNVATLPAE KMKILELPFASGDLSMLVLLPDEVSDLER
  IEKTINFEKLTEWTNPNTMEKRRVKVYLPQMKIEEKYNLTS
  VLMALGMTDLFIPSANLTGISSAESLKISQAVHGAFMELSEDGIEMAGST
  GVIEDIKHSPESEQFRADHP FLFLIKHNPTNTIVYFGRYWSP
• 1 qikdllvsss tdldttlvlv naiyfkgmwk tafnaedtre
  mpfhvtkqes kpvqmmcmnn
• 61 sfnvatlpae kmkilelpfa sgdlsmlvll pdevsdleri
  ektinfeklt ewtnpntmek 121 rrvkvylpqm kieekynlts
  vlmalgmtdl fipsanltgi ssaeslkisq avhgafmels
• 181 edgiemagst gviedikhsp eseqfradhp
  flflikhnpt ntivyfgryw sp
• Identifiers
• accession, accession.version or gi's
• e.g., p01013, AAA68881.1, 129295, gi129295
• FASTA format

GenBank format
5
Query Sequence in FASTA Format

• FASTA definition line ("def line") that begins
  with a gt, followed by some text that briefly
  describes the query sequence on a single line
• Up to 80 nucleotide bases or amino acids per line
• Blank lines not allowed in the middle
• Example
• gtgi129295spP01013OVAX_CHICK GENE X PROTEIN
  (OVALBUMIN-RELATED) QIKDLLVSSSTDLDTTLVLVNAIYFKGMWK
  TAFNAEDTREMPFHVTKQESKPVQMMCMNNSFNVATLPAE
  KMKILELPFASGDLSMLVLLPDEVSDLERIEKTINFEKLTEWTNPNTMEK
  RRVKVYLPQMKIEEKYNLTS VLMALGMTDLFIPSANLTGISSAESLKIS
  QAVHGAFMELSEDGIEMAGSTGVIEDIKHSPESEQFRADHP
  FLFLIKHNPTNTIVYFGRYWSP
• Additional information

6
What does BLAST tell you?

• Putative identity and function of your query
  sequence
• Helps to direct experimental design to prove the
  function
• Find similar sequences in model organisms (e.g.,
  yeast, C. elegans, mouse), which can be used to
  further study the gene
• Compare complete genomes against each other to
  identify similarities and differences among
  organisms

7
Variety of BLASTs
http//www.ncbi.nlm.nih.gov/BLAST/
8
BLAST Programs Which One to Use?

• Depends on
• What type of query sequence you have (nucleotide
  or protein)
• What type of database you will search against
  (nucleotide or protein)
• BLAST program descriptions
• brief list
• BLAST program selection guide

9
Commonly Used BLAST Programs

• Examples of BLAST programs
• BLASTN
• Nucleic acids against nucleic acids
• BLASTP
• Protein query against protein database
• Usually better to use than nucleotide-nucleotide
  BLAST
• Since the genetic code is degenerate, blastn can
  often give less specific results than blastp
• ...but... what if we don't have a protein query
  sequence. What are our options?
• BLASTX
• Translated nucleic acids against protein database
• One way to do a protein BLAST search if you have
  a nucleotide query sequence
• The BLAST program does the translating for you,
  in all 6 reading frames

10
BLAST Databases Which One to Search?

• What type of data do you want to search against?
  For example
• Characterized sequences?
• Specialized sequences?
• Complete genomes or chromosomes?
• BLAST database descriptions are available in the
  
• BLAST help document
• BLAST program selection guide

11
Request ID RID

• An RID is like a ticket number that allows you to
  retrieve your search results and format them in
  many different ways over the next 24 hours.
• If you've saved RIDs from your recent searches,
  you can enter the RIDs directly using the
  Retrieve results with a Request ID page, which is
  accessible from the bottom of the BLAST home page

12
Search Results Understanding the Output

• Reference to BLAST paper
• Reminders about your specific query
• RID
• query sequence reminder (contains the information
  from your FASTA def line)
• what database you searched against
• Graphical summary
• shows where the hits aligned to your query
• colors indicate score range
• mouse over a colored bar to see info about that
  hit
• Text summary (GI numbers and Def lines)
• GI links to complete record in Entrez
• Score links to pairwise alignment between your
  query sequence and the hit
• Pairwise alignments
• BLAST statistics for your search

13
Database Search w/ BLAST

• Finding similar sequences is a primary use of
  bioinformatics
• BLAST

Enter sequence Choose DB Hit
Acknowledgement Slides 15 19 are adapted from
lecture notes of Professor Chau-Wen Tseng of CS
Department at the University of Maryland with
permission.
14
Database Search w/ BLAST

• Versions of BLAST
• BLASTN
• Nucleic acids against nucleic acids
• BLASTP
• Protein query against protein database
• BLASTX
• Translated nucleic acids against protein database
• TBLAST
• Protein query against translated nucleic acid
  database
• TBLASTX
• Translated nucleic acids against translated
  nucleic acids

15
Database Search w/ BLAST
16
Database Search w/ BLAST

• BLAST graphic result

17
Database Search w/ BLAST

• BLAST result
• 0Matching sequences w/ bit-score E-value
• 0Hyperlinks to database entry for sequence
• Example
• gi17330420gbBH384278.1BH384278 ... 153 3e-36
• gi17320126gbBH373984.1BH373984 ... 140 9e-34
• gi17338337gbBH392196.1BH392196 ... 112 8e-25
• gi20373967gbBH771010.1BH771010 ... 105 1e-21
• gi17314411gbBH368367.1BH368367 ... 104 2e-21
• gi17332712gbBH386570.1BH386570 ... 64 3e-21
• Hyperlinks to sequences Bit Score
  E-value

18
BLAST Statistical Evaluation

• E Value
• The number of different alignments with scores
  equivalent to or better than alignment score that
  are expected to occur in a database search by
  chance.
• The lower the E value, the more significant the
  score.

19
BLAST How It Works

• Find high scoring local alignments between query
  sequence and target database
• Assumption
• True match alignments very likely to contain
  within them very high scoring matches
• Steps
• Seeding
• Searching
• Extension
• Evaluation

20
BLAST Steps

• Seeding
• For each word of length w in the query (w-mer),
  generate a list of all possible words (neighbors)
  with a score of at least threshold T (determined
  by using the scoring matrix)
• Default
• w 3 for protein
• w 11 for DNA
• BLASTn, however, does not find neighbors. It
  uses the 11-mers from the query only.

21
Query word (w 3)
Query GSDFWQETRASFGCSLAALLNKCKTPQGQRLVNQWIKQPLMDK
NRIEERLNLVEAFGCATSWPI
PQG 18 PEG 15 PRG 14 PKG 14 PNG 13 PDG 13 PHG 13 P
MG 13 PSG 13 PQA 12 PQN 12
Neighborhood words
Neighborhood score threshold (T 13)
This example uses BLOSUM 62.
22
BLOSUM 62
23
BLAST Steps

• Searching
• Determine the locations of all common words
  between the query and the database (word hits)
• Identifies all word hits

24
Query word (w 3)
Query GSDFWQETRASFGCSLAALLNKCKTPQGQRLVNQWIKQPLMDK
NRIEERLNLVEAFGCATSWPI
PQG 18 PEG 15 PRG 14 PKG 14 PNG 13 PDG 13 PHG 13 P
MG 13 PSG 13 PQA 12 PQN 12
Neighborhood words
Neighborhood score threshold (T 13)
Hit
Query SLAALLNKCKTPQGQRLVNQWIKQPLMDKNRIEERLNLVEA
Subject TLASVLDCTVTPMGSRMLKRWLHMPVRDTRVLLERQQT
IGA
25
Implemention of BLAST Steps 1 2

• (Using DNA as an example)
• Given a query sequence q, a database sequence d,
  and the word size w, the following steps are
  needed
• Find w-mers from q and create a table (such as
  hash table) to store them and record the position
  of their occurrences.
• Scan d to find if a w-mer in d is also in q,
  i.e., in the table.

26
Implemention of BLAST Steps 1 2
Example Suppose q ACACAT, d GTACACGTT,
and w 3.   Build the table The 3-mers of q are
ACA, CAC, ACA, and CAT. Their positions are
stored in the table. The position of a sequence
starts at 1.  
4
1
3
2
27
Implemention of BLAST Steps 1 2
Example (contd) Match w-mers Use the 3-mers in d
one at a time to find its position(s) in q using
the hash table. Here is the result GTA not
found TAC not found ACA 1, 3 CAC 2 ACG not
found CGT not found GTT not found
4
1
3
2
28
Implemention of BLAST Steps 1 2

• Questions
• How many w-mers are there in q?
• How many w-mers are there in d?
• What is the time required to build the table?
• What is the time required to find all the
  positions?
• An alternative is building table using w-mers in
  d and then scan q to find positions of the
  w-mers. Why is it not recommended?

29
BLAST Steps

• Extension
• Extend hits to find HSPs (high-scoring segment
  pairs) that have scores higher than a threshold
• Introduce gaps using dynamic programming
• Problem of extension
• Time-consuming to find the highest score
• Solution (heuristic)
• Extend until score drops a value of X

Example ABCDEFGHIJKLMNOPQRST
ABCDEFZYIJKLMXWVUTAB
1234565456789876565 ? Score
00000012100001234345 ? Drop off score
Match 1 Mismatch -1 X 5
30
Query word (W 3)
Query GSDFWQETRASFGCSLAALLNKCKTPQGQRLVNQWIKQPLMDK
NRIEERLNLVEAFGCATSWPI
PQG 18 PEG 15 PRG 14 PKG 14 PNG 13 PDG 13 PHG 13 P
MG 13 PSG 13 PQA 12 PQN 12
Neighborhood words
Neighborhood score threshold (T 13)
Hit
Query SLAALLNKCKTPQGQRLVNQWIKQPLMDKNRIEERLNLVEA
LAL TP G R W P D ER
A Subject TLASVLDCTVTPMGSRMLKRWLHMPVRDTRVLLERQQT
IGA
31
BLAST Steps

• Evaluation
• Select all HSPs from step 3 having a score of at
  least S (for some parameter S). The value of S
  should be large enough so that unrelated
  sequences shouldnt have HSPs with scores S.
• Evaluate the statistical significance of extended
  hits
• Report only those above the determined threshold

32
BLAST Statistical Evaluation

• For local, ungapped alignments
• m size of query
• n size of database
• E expected of HSPs with scores at least S
• p prob of finding at least one HSP with S

• good tutorial at
• http//www.ncbi.nlm.nih.gov/BLAST/tutorial/Altschu
  l-1.html

33
Interpretations of Expected Value

• Expected value ranges
• E lt 10-100 ? very low, homologs or identical
  genes
• E lt 10-3 ? moderate, may be related genes
• E gt 1 ? high, probably / may be unrelated
• 0 0.5 lt E lt 1 ? ??? In the twilight zone Try
  detailed search
• If database search
• Long list of gradually declining of E values ?
  large gene family
• Long regions of moderate similarity ? more
  significant than short regions of high identity
• Biological relevance
• Still need to determine biological significance!!!

34
Other BLAST Algorithms

• BLAST2
• Gapped-BLAST
• PSI-BLAST
• BLAT
• Mega BLAST
• etc.