Transcript Using genome browsers

Using genome browsers
Visualization and data
repositories
Motivation
Aside from R, genome browsers are
arguably the most important tool in
computational genomics
…but is not widely used in the
experimental community
The browser gives you an immediate
edge - you can look at data, form
hypothesis and up-and download data!
In this course
1: How to use the web interface;
understanding the data types
2: How to download and upload data to
the browser; interaction with R
3: How to make complex analyses
between data types; Galaxy and R
Today's teaching:
• Lectures with genome browser examples
• Short discussions with your neighbour
• Exercises
Kick starting with a challenge
• You are a major sequencing center
• You have sequenced the killer whale
(Orca) genome - you have the whole
genome as a stretch of ACGTs
• How do you make sense of this and
show it to others? What value does the
data have in itself?
• 2 minutes with your neighbour
Jim Kent, assembly-guru.
Some profound words about
the genome sequence
“Well, it has a lot of G, C, A and Ts”
Genomes are worthless
• …without any annotation
• What type of annotations do we want to
put on genomes?
• 2 minutes with your neighbour
Examples:
• 'DNA' annotation:
– Known genes
– Predicted genes
– Repeats, transposons, CpG islands
– Conservation across species
• 'Dynamic' annotation:
– Known transcripts
– Expression data
– DNA modifications
How to present this data?
• Plain text files are useless..for most
biologists
• Use the genome sequence as a frame,
on which we map real data or
predictions
The idea of the browser
• Based on the genome, we can
– Zoom up and down, and scroll sideways
– See the data in different representations
– Select WHAT data we want to see (way to much
data to look at all at once)
• Important side-effect: if we map all interesting
data, it means that all data is at one place,
which means that we can download what we
are interested in to do analysis!
The three browsers
• UCSC genome browser
– http://genome.ucsc.edu
– Updated often, simple but powerful interface. Very
simple underlying data formats
• ensEMBL
– http://www.ensembl.org
– More complex web interface, with multiple zoom
levels. Very complex underlying data formats
• The generic genome browser
– http://www.gmod.org/GBrowse
– Actually more a software development platform, so
that you can do your own. Resembles UCSC more
than ensEMBL
In this course…
• We will only use the UCSC browser due to
– Simplicity
– Lecturer bias
– The galaxy tool - a very nifty web-tool to do power
user analysis on UCSC data (more later)
• If you know this browser, other browsers are
easy to understand
Basic concepts
• Zooming
• Data tracks
Data tracks -the problem
Example: The road from Melby to Ølby
Melby
Ølby
5 km
10 km
Melby
Ølby
5 km
10 km
Data tracks -the problem
Example: The road from Melby to Ølby
Melby
Ølby
5 km
10 km
Melby
Ølby
5 km
10 km
Melby
Ølby
5 km
10 km
Data tracks -the solution
Melby
Ølby
5 km
houses
trees
Monday
Sunday
5 km
10 km
This is how genome browsers
show the data
Chromosome
position
Gene track
mRNA track
Exons
Introns
Annotation tracks
• A track is often one source of data, from a
particular place, that is mapped to the
genome
• Data can be viewed as “blocks” with a start
and an end, expressed as chromosome
coordinates
• It is important to know what the data is before
trying to interpret it
• We will first look at the “human mRNA” track
Human mRNA track
• What the guys at UCSC did:
– Take all the known mRNAs in Genbank,
and map these to the human genome
using a software called BLAT (similar to
blast). Everything that hits will be shown in
this track.
– What is the pros and cons of this
approach? What are the limitations? 2
minutes with your neighbour!
Example answers:
Pros
Simple, and no filtering - leaving me to make
interpretation
Cons
Not real annotation - again, leaving me to
make interpretation
Heavily reliant on the data source quality
Limited by the extent of data
A short non-interactive tour
• We will use the browser extensively
from now on
• But first, I will guide through a few key
concepts - otherwise confusion ensues
when trying the real thing
What version of the genome
do you have?
• Genome sequences are based on many short
sequenced reads, which then are assembled
into a single sequence
• This is very tricky, and we get slightly updated
genomes at regular intervals
• A version of the genome is called an
assembly
• So, whenever you say that you are using a
genome sequence to do something, you have
to say what assembly you are working on!
More about assemblies
• The official naming system is
– [species abbreviation][assembly number]
For instance hg17 (human nr 17), or mm8
(mus musculus 8)
There is an alternative way: the date of the
release.
So, hg17 is also called “Human May 2004”
Even more about assemblies
Rules of thumb:
The newer an assembly, the “better”
Some older assemblies have more data
mapped to them (because they have been
around longer)
Some genomes are new, and unstable: updates
come often, and big differences between
updates. Some are more mature (like human)
Selecting species & assembly
Species
Where on the genome
Assembly: the genome “version”.
Looking at the genome,
with mRNAs
Chromosome overview
Different mRNAs (same gene)
Direction of arrows
shows strand
Zooming in
(We'll learn how later)
Some points:
•Transcription, in this case, is right to left transcription on the minus strand - shown by the
arrows
•Two of the mRNAs start here, the others start even
further upstream. Probably alternative promoters
•The fat, two-colored blocks are predicted to be
protein-coding parts
Note that
•There are parts of mRNAs that are not translated so called UTRs
•There is one mRNA that is clearly non-coding (might
have a stop-coding further upstream)
Zooming even further down - we see the actual DNA
Codons
Clicking on any of these mRNAs take you to the
corresponding Genbank entry
Different data representations
Each data track has a selection 'box'
Use this to :
-turn tracks on or off
-change visualization
Full
examples
Squished
Dense
Time to try it out..
• Important: the genome browser shows many
tracks by default, some which are named in
a confusing way
• Don’t let this throw you. We will walk them
through!
• Goto http://genome.ucsc.edu/
• Click 'Genome browser' to the left
We'll use default position for now,
so just click the 'Submit' button
(which is on the right)
Overwhelmed?
Many types of data! We will only use
some, others you can explore
yourselves
Below the image, the data tracks are
categorized for easier access:
Let’s look only at the Human mRNA
track as before
Challenge:
Turn off all tracks, except “base position”
and “human mRNA”!
(Expand/collapse the categories, then hide tracks.
Use 'refresh' to update the image.)
Challenge
Using the following buttons, and what we already went through, find out:
What is the DNA sequence of the first two codons of mRNA DQ892408?
What is the “gene name” of the mRNAs we are looking at?
Are the two longest RNAs starting at exactly the same place?
What are the neighboring genes?
Before we go any further…
What are all these data? What can we use them for?
Fast info on a given track:
•
•
•
•
Click on the actual track name (over the box)
What does the “refseq genes” track hold?
What is the difference to “other refseq” or “Genscan genes”
When would you use each track?
• It is not realistic to go through all tracks in this
course
• …and not meaningful, because new tracks
are added over time
• We will go over the main types of tracks, and
the relevant experimental methods for
producing the tracks
• Understanding what we are looking is very
necessary for meaningful interpretation
Big groups of things,
summarized
• Sequence features
– CpG islands
– Repeats
• Transcripts or part of transcripts
– mRNA, ESTs
• The so-called genes (predicted or experimental)
• Tiling array expression data
• Chip-Chip
• Variation within species (SNPs)
• Conservation and alignments between species
– net alignments, Phastcons scores,
• The ENCODE dataset
Between transcription and translation
– the modern RNA world
• After transcription, RNAs are
immature (precursor mRNAs).
Processing RNAs give mature
mRNAs, which gives access to the
cytoplasm, and translation. As
usual, we know only a small part of
the mechanisms...
• 5' CAP structure is added
• 3' polyA stretch is added
• Splicing (not always!)
• RNA editing (rare?)
Splicing
Problem:
We want to know what mRNA look like...
but RNA is unstable, can't be sequenced directly
Solution:
Turn them into cDNA first.
Into a plasmid – so, we have a library of plasmids each carrying one cDNA
This is a “cDNA library” that later can be sequenced or used for other things
General problems with cDNA
sequencing:
• Reverse transcriptase falls off
• Hard to sequence long transcripts
• Many cDNAs are identical
– Very expensive if you want to sequence all
unique molecules
Solving the problem
Only sequence parts of cDNAs - these are
called ESTs(more in a few slides)
Semi-recent development: sequencing of
full-length cDNAs, using
– Cap-trapping
– PolyA primers
– subtraction
Subtraction: how to only get
RNAs you have not seen yet
• Simple concept:
• For a cDNA sample, we add an excess
of abundant RNAs. These will hybridize
• Then, we remove everything which
hybridized
• …and sequence the rest
Discuss with your neighbour
(2 min)
Say that we have two cDNA libraries - one is
subtracted, one is not
What are they good for?
Expression (how many transcripts of a
certain gene)?
Annotation and gene discovery?
Visualizing and annotating
cDNAs in the genome browser
• The genome is actually needed to make sense of cDNAs,
especially if it is not protein-coding
• A general approach is to map your cDNA to the genome
using an alignment algorithms
• Here, we will use BLAT and the UCSC browser
• Should be straight-forward, but...lets try it out: See the
course page for 3 mouse sequences in the blat_seqs file
– I will do one in real-time
• Assume these are new sequences that you must say
whether they are good enough to be part of the genome
browser
Bottom line
• cDNA <->genome is sometimes trivial, but
can become very tricky. Bear this in mind
when you look at genome mappings – this
is the process they are annotated with!
• cDNAs are often good quality, but always
be sceptical unless there are multiple lines
of evidence
• Biological knowledge helps here – sanity
checks become easier
More on the problem of
sequencing cDNAs
Hard to sequence full-length cDNAs
…and expensive to sequence many
If we cannot sequence the whole cDNAs…
Only sequence parts of cDNAs - these are
called expressed sequence tags: ESTs
Expressed sequence tags (EST)
Cheaper, and easier to scale
up
Problems:
many ESTs are simply trash –
the result of over-enthusiastic
sequencing
For longer genes, no
coverage of the middle part
Complementary information to
cDNAs
• Can be used for expression studies
(more later)
• Many MORE of them than full-length
cDNAs - higher coverage
• If you only have ONE cDNA for a given
isoform, ESTs can help to “validate it”
So-called “gene” tracks
• We have now seen that often a “gene” have
many mRNAs - forming a “transcription unit”
• If you have many mRNAs, it is good to have
summary tracks of genes or transcription
units
• The UCSC browser has (at least) two of
these:
– The RefSeq track
– The “Known genes” track
Refseq
• Refseq is actually database with high-quality
cDNAs, from NCBI. So, a Refseq sequence
always has at least one identical cDNA in
GenBank.
• Good, because some individual cDNAs are
trash, and we get a more manageble dataset
• Bad, because the criteria used are somewhat
arnitrary. For example, “long cDNAs are
better than short”
Known Genes
A track made by the UCSC people, which
uses multiple databases (Refseq,
uniprot, etc)
Horrible name - easy to misunderstand it it is NOT all known genes!
If clicking on individual genes, you get
very nice summaries, sometimes with
expression information
Searching by gene name
• If you put in a gene name, or an accession
number in the coordinate box, the browser
will search the mRNA, Refseq and Known
Genes tracks (and some more) for this name,
and give you a list if you get more than one
hit
• Is usually easy: here is an example: the
Dicer1 gene (an important RNAse)
CpG islands
A CpG dinucleotide is simply a C followed by a G
CpGs are uncommon (1%) in vertebrate genomes, due
to that the C in the CG is easily methylated and then
deaminated into a T
However, there are stretches of CpG rich dinucleotides,
called CpG islands
These are correlated with promoters - around 50% of
promoters have a CpG island. Function is unclear!
In the UCSC browser, this is simply called the CpG
island track
Repeats
Large portions of the genomes are “repeats”,
classified into two main types:
1)Tandem repeats
Two or more nucleotides are repeated,
directly after each other
ATTCGATTCGATTCG
(number of repeats are used in crime forensics
and parentage tests)
2) Interspersed repeats
Results of RNA-mediated transposition (not in
this course)
Repeats, cont
• Generally, repeats are considered
“uninformative”, and presents problems
when aligning things to the genome
• However, there are clear cases of
functional repeats
• In the UCSC browser, all repeats can be
turned on in the repeat track
Lets look at these things
• 5 minutes with your neighbour:
• Look at the RPS9 gene, and turn on Refseqs,
known genes, human mRNAs, ESTs, CpG
islands and repeats
• How well does refseqs, ESTs and Known
genes correlate
• Are there any CpGs or repeats - where are
they located? What type of repeats are there?