1. Lab 4.1: Multiple Sequence Alignment & Phylogenetic Trees
February, 2006
Jennifer Gardy
Centre for Microbial Diseases & Immunity Research
University of British Columbia
Lab 4.1
(c) 2006 CGDN

2.
Lab 4.1

3. Goals
Learn how to create, view, and interpret multiple sequence alignments and phylogenetic trees
Understand the difference between orthologs and paralogs
Use an alignment and a tree to deduce information about biological sequences
Complete the phylogeny assignment
Complete Module 3 of the Integrated Assignment
Lab 4.1

4. Outline Quick review of MSA How Clustal works Research Question
Lab 4.1
February, 2006
Quick review of MSA
How Clustal works
Research Question
Worked example 1: Creating an MSA
Neighbour-joining trees
Bootstrapping
Worked example 2: Creating/viewing a tree
Free time to work on phylogeny assignment
Lab 4.1
(c) 2006 CGDN

5. MSAs: A Quick Review Why perform an MSA? How does one perform an MSA?
Visualize trends between homologous sequences:
Shared regions of homology
Regions unique to a sequence within a family
Structural/functional motif
As the first step in a phylogenetic analysis
Improve accuracy of structure predictions
How does one perform an MSA?
Automated alignment with manual editing
Considerations:
Select sequences carefully:
Homologous over length
No unrelated sequences
Computational intensity
Lab 4.1

6. Clustal – A Common MSA Tool
Progressive alignment strategy:
Sequences used to make guide tree
Most similar two sequences aligned = consensus
Next closest sequence aligned to the consensus
13 MITTEN
1 MITTENS
2 KITTIES
3 SMITTEN
4 KITTENS 1 3 4 2
1 -MITTENS
3 SMITTEN-
13 MITTEN
4 KITTENS
134 ITTEN
Manual editing:
Fine adjustment of particular columns
Incorporate specific knowledge
Removal of gappy bits
Important for phylogenetic analysis
Removal of parts of/whole sequences
Non-homologous regions
Sequences included by error
134 ITT-E
2 KITTIES
Lab 4.1

7. Research Question: Background
Olfaction: our sense of smell
Small chemical compound (odorant) binds olfactory receptor (OR) on nasal epithelium
OR changes shape, signaling cascade perception of odor
>1000 OR genes in mammals (~350 active in humans), each binds 1 or more odorants
But we smell many more scents than this! How?
Combinatorial effects of Ors
Odorants are enantiomers:
Non-superimposable mirror images
Each enantiomer has unique scent
Lab 4.1

8. Research Question: Background
ORs are G-protein coupled receptors (GPCRs)
All are very similar in structure
7 transmembrane helices
3rd transmembrane helix thought to be important for odorant specificity
Odorant:OR relationship not known for most of the OR genes
Starting to employ bioinformatics techniques
Lab 4.1

9. Research Question: Outline
Computational analysis of OR sequences:
4 human sequences: OR6N1, OR6N2, OR6K2, OR6K6
OR6N1 binds R(-) carvone (spearmint)
OR6K2 binds R(+) limonene (citrus fruit)
3 mouse homologs: olfr420, olfr425, olfr429
Can we figure out the odorant specificity of the other odorant receptors?
Which residues might be important for odorant binding specificity?
Lab 4.1

10. Starting up ClustalX Day 4 wiki > OR_proteins.txt
Open nano or other text ed., paste sequence in and save
Start ClustalX - $ clustalx
File:
Load sequences
Edit:
-Remove all gaps
Alignment:
-Do complete
alignment
-Alignment
parameters
Trees:
-Bootstrapped NJ
-Output format
options
Name
Window
Sequence
Lab 4.1

11. OR Proteins in ClustalX
File > Load sequences > OR_proteins.txt
How are unaligned sequences displayed?
Left-aligned, in order of input
Default colouring (identity) – see help file for details
Conservation score graph
Lab 4.1

12. Alignment Parameters
Alignment > Alignment Parameters > Multiple Alignment Parameters
What effect might increasing the gap penalties have on alignment?
Decreasing the gap penalties?
Increased gap penalties:
Fewer gaps
May cause program to miss important insertion/deletion events
Decreased gap penalties
More gaps
Makes it easier to align sequences that may not be closely related to each other
Lab 4.1

13. Do an Alignment Alignment > Do complete alignment
Generates an .aln file
Lab 4.1

14. Building an NJ Tree - An Example
Cbw protein from cat, rat, bat, mat and Matt
Compare all sequences to each other.
Assign divergence values to each pair
Assemble the values in a distance matrix
Cat
Rat
Bat
Mat -
0.7
0.8
0.2
1.0
Matt
0.6
0.4
0.5
0.9
Lab 4.1

15. Building an NJ Tree 4. Arrange the subjects in a “star” phylogeny
Lab 4.1

16. Building an NJ Tree 5. Fuse the two branches with the least divergence
Cat
Rat
Bat
Mat -
0.7
0.8
0.2
1.0
Matt
0.6
0.4
0.5
0.9
Lab 4.1

17. Building an NJ Tree
6. Create a new distance matrix using the fusion consensus sequence
Cat
RatBat
Mat -
0.75
1.0
0.8
Matt
0.6
0.45
0.9
7. Fuse the next two closest sequences
8. Repeat until tree completed
Lab 4.1

18. A Completed Tree
Alternative displays are possible:
Lab 4.1

19. A Word about Bootstrapping
1001 definitions, none of which have to do with boots. Or straps.
In phylogenetic analysis, bootstrapping is a simple test of phylogenetic accuracy:
Does my whole dataset strongly support my tree? Or was this tree just marginally better than the other alternatives?
Lab 4.1

20. Bootstrapping – The Wordy Version
Original dataset is “randomly sampled with replacement”
Multiple (N=100, 1000, etc…) “pseudo-datasets” of the same size as the original are created
Each of the N pseudo-datasets is used to create a tree
If a specific branching order is found in X of the N trees, that node is given the bootstrap support value X
X values of 70% or more = very reliable groupings
Lab 4.1

21. Bootstrapping – The Picture Version
Slice original MSA of Y residues into Y columns, put the columns into a hat
Pull out a random column, place it in column #1 of your new test set
3. Put the column back in the hat
4. Pull another column from the hat, place it in column #2 in the test set, put it back
5. Repeat until a pseudo-dataset of Y columns has been made
“random sampling”
“with replacement”
Lab 4.1

22. Bootstrapping Repeat N number of times to generate N pseudo-datasets
For each pseudo-dataset, draw a tree (yields N trees)
Compare your tree to all N trees. How often do the branching orders in your tree appear in the N pseudo-trees? 3
On branches of your tree, write # of times that branch appeared in your pseudo-dataset trees 2 1 2
Lab 4.1

23. Our Bootstrapped Neighbour-Joining Tree
Trees >
“Exclude positions w/ gaps”
Any column with 1+ gaps
Deletes uninformative regions
Not good for gappy MSAs
“Correct for multiple subs.”
M -> V -> L -> V
3, not 1, mutations
Correction formula makes distances proportional to time since divergence
Trees > Output Format Options
Bootstrap labels on “node”, not “branch”
Makes for easier visualization
Bootstrap N-J Tree
Lab 4.1

24. Draw an Bootstrapped NJ Tree
Trees > Bootstrap NJ Tree
What does this file look like? (
OR6K2: ,
olfr420: )
: ,
OR6N1: ,
olfr429: )
: ,
OR6N2: )
: )
: ,
OR6K6: ,
olfr425: );
Not very tree-like
Lab 4.1

25. View the Tree with NJPlot
Lab 4.1
February, 2006
NJPlot – very basic
njplot at prompt
Turn on bootstrap value display
Can swap nodes
Many other tree viewing programs available
Treeview, 3 different views:
Lab 4.1
(c) 2006 CGDN

26. Remainder of lab time/Evening open lab
Complete the phylogeny assignment!
View tree on-screen
Use tree and facts on pg. 3,4 of Lab 4.1. notes to answer Q1-Q5
Look at alignment of single transmembrane segment
TM helix 3, thought to be important for odorant specificity
Answer remaining assignment questions using this data
Attention biologists! Karma alert! Help your teammates to understand evolution today, and they’ll help you understand programming tomorrow!
Module 3 of the Integrated Assignment
MSA/tree analysis of sequences found in Module 2
Last IA component for Week 1.
Phylogeny assignment due 9am, IA due 11am Saturday
Half-day open lab tomorrow afternoon, plus evening times
Lab 4.1