Biology 301 Computational Biology HomologySimilarity, The Matrices, BLAST Bioinformatics Ch. 8 secti

1
Biology 301 - Computational BiologyHomology/Simil
arity, The Matrices, BLAST Bioinformatics Ch. 8
section MEP Ch. 2-3 Blast Tutorial
2
General Introduction

• Introduction to terms
• Similarity same monomers
• Homology inference, predicts phylogeny
• BFD, like many molecular biologists, improperly
  bandies around these terms. Which are "results"
  and which are "discussion? What did you think
  of the genome papers use of these data and terms?
  

3

• Compared molecules SHOULD
• Have a common ancestor
• Have a similar function
• Be homologous at ALL levels
• - enough similarity to be aligned
• - enough variability to be informative
• EACH monomer and domain one trait
• THEREFORE

4
(No Transcript)
5

• Alignment - what we are building up to
• Lining up homologous monomers/domains
• Every alignment decision assumption
• Known functional or structural domains
• Computer programs can only deal with simple,
  predictable patterns - similarity. Ultimately,
  the researcher has to examine computer
  alignments, integrating biological information
  determined from the literature or bench.

6

• Substitute
• Hemoglobin from each of taxa
• Thermal stability
• Enhanced O2-binding
• Decreased CO-binding
• Binds 4 peptides
• Binds 2 peptides

7
Thermal, O2-binding, CO-binding, 4 peptides, 2
peptides
8
(No Transcript)
9

• Alignments compute
• S substitutions/mutations
• G length of gaps (insertions/deletions)
• W gap penalty (VERY subjective)
• Computer programs use different matrices for
  weighing (or not weighing) substitutions.
  Sometimes, you can control them, sometimes they
  are default

10
e.g. Dayhoffs Amino Acid Matrix -
PAMsubsitutions occur more often between amino
acids that are similar in terms of biochemistry

• General Characteristics of PAM
• PAM accepted point mutations
• Based on changes/100 aa (eukaryotic)
• Table units likelihood of change
• Some aas seldom change C, G, W
• High values similar (A/G 21 A/K 4)
• Studying microbial and mitochondrial proteins
  has lead to other matrices.

11
e.g. Contrast Two Nucleotide Matrices

• Jukes/Cantor Model
• All nucleotides equal, typically all 1
• Kimuras Two-Parameter Model
• Transitions more frequent than transversions
• Scores can vary but typically T 2 t 1
• Many WAY complicated matrices have been
  developed for nucleotides more later.

12
Basic Local Alignment and Search ToolRapid
alignment and similarity program.

• In an nutshell
• Allows you to input any "unknown" sequence
• Compares unknown to ENTIRE database
• Outputs ranked hitlist with scores and links

13

• Things BLAST is useful for
• Finding CDS in a large genome segment
• Predicting protein function - Hmmm
• Predicting protein structure - Hmmm
• Finding gene or protein family members
• Checking for sequencing errors
• BUT - "BLAST hits are not transitive, unless the
  alignments are overlapping"

14

• SIMPLE explanation of how BLAST aligns
• Assumes all sequences, lengths AVERAGE
• EXTREME gap penalty (called gap cost)
• Often results in SHORT regions aligned
• PAM or Equal Weighting
• FAST and POWERFUL but not intricate
• No "biology" used for nucleotides
• Some versions back- or forward-translates every
  possible frame for comparison.

15

• Overview of BLAST output - 35 pages
• Graphic - clickable color-coded similarity
• Hitlist - ranked names/accessions, scores
• Alignments - EVERY used query/hit
• Parameters - list of parameters used/selected

16
(No Transcript)
17

• The Hitlist Scores
• Bit Scores statistical significance
• Less than 50 is considered bad
• E-value probability alignment due to chance
• Keep in mind - these scores apply ONLY to the
  OFTEN-SHORT alignments used! Read Box p. 229 and
  make sure you actually LOOK at your alignments!

18

• BLAST Alignments
• Identity within aligned region, same
• 25 or higher is considered good
• Length the actual length of the alignment

19

• BLASTing DNA
• BLASTn DNADNA db (70 similar)
• TBLASTx translated DNADNA db (families)
• BLASTx translated DNAprotein db (CDS)
• BLASTing Proteins
• BLASTp - aa sequenceprotein db (domains)
• TBLASTn - aa sequenceDNA db (families)
• Families - goal is research and discovery of new
  relationships.

20

• Changing BLAST parameters
• Direct searching some db or db members
• MASK some regions (conserved, repetitive)
• Change gap penalties
• Change substitution matrix
• All of these changes alter output significantly
  and it is IMPORTANT to record changes in
  procedures. In general, you should NOT change
  BLAST parameters in this course!