Large-scale biology projects such as the sequencing of the human genome and gene expression surveys using RNA-seq, microarrays and other technologies have created a wealth of data for biologists. However, the challenge facing scientists is analyzing and even accessing these data to extract useful information pertaining to the system being studied. This course focuses on employing existing bioinformatic resources – mainly web-based programs and databases – to access the wealth of data to answer questions relevant to the average biologist, and is highly hands-on. Topics covered include multiple sequence alignments, phylogenetics, gene expression data analysis, and protein interaction networks, in two separate parts. The first part, Bioinformatic Methods I, dealt with databases, Blast, multiple sequence alignments, phylogenetics, selection analysis and metagenomics. This, the second part, Bioinformatic Methods II, will cover motif searching, protein-protein interactions, structural bioinformatics, gene expression data analysis, and cis-element predictions. This pair of courses is useful to any student considering graduate school in the biological sciences, as well as students considering molecular medicine. These courses are based on one taught at the University of Toronto to upper-level undergraduates who have some understanding of basic molecular biology. If you're not familiar with this, something like https://learn.saylor.org/course/bio101 might be helpful. No programming is required for this course although some command line work (though within a web browser) occurs in the 5th module. Bioinformatic Methods II is regularly updated, and was last updated for March 2019.
Lab Discussion
Loading...
来自 University of Toronto 的课程
Bioinformatic Methods II
197 个评分
从本节课中
Cis Regulatory Systems
When and where genes are expressed in tissues or cells is one of the main determinants of what makes that tissue or cell the way it is, both in terms of morphology and in terms of response to external stimuli. Gene expression is controlled in part by the presence of short sequences in the promoters (and other parts) of genes, called cis-elements, which permit transcription factors and other regulatory proteins to bind to direct the patterns of expression in certain tissues or cells or in response to environmental stimuli: We'll explore a couple of sets of promoters of genes that are coexpressed with AP3 from Arabidopsis, and with INSULIN from human, for the presence of known cis-elements, and we'll also try to predict some new ones using a couple of different methods.
与讲师见面
Nicholas James ProvartProfessor
Cell & Systems Biology
Right, in this week's lab we're exploring cis elements, and cis elements are really
important for the regulation of the expression level of a given gene.
So they determine where and when a given gene is active in particular tissues or
in cell types or in response to a perturbation / environmental stress.
So what we're looking at in question A is the output of a word
count program which looks at the frequency of occurrence of particular k-mers.
We take all the promoters,
in this case APETALA3 coexpressed genes, promoters of those genes,
and we look for whether particular k-mers, 6-mers in this case,
are over represented in those promoters relative to random sets of promoters.
And question a asks what the colour of the background distribution is and
that colour is in blue, here.
So this is just randomly picked promoter sets and
the colour for the distribution of the input cluster, i.e. our AP3
coexpressed promoters, genes that are coexpressed with AP3 -
promoters thereof, the distribution is shown in red, here.
So in this case, we're looking at several k-mers, 6-mers here.
Each row consists of two distributions,
the blue distribution and the red distribution.
And we see gradually, decreasing
distances between those two distributions as the significance decreases.
So these first two are quite significant and
in fact the first two do appear that they could overlap.
So, if we look at the composition of the first k-mer it's TAATTA.
And the second one is ATTAAT and
the first four nucleotides of the second six-mer
could overlap with the last four nucleotides of the first one.
It seems like they could overlap.
All right, so question c asks, are the two top 6-mers similar to any known motif?
And if we look at, here's the second one actually,
we see that it's similar to AGAMOUS binding sequence
from Arabidopsis thaliana, as is the first one, here.
So again, AGAMOUS pops up as a consensus binding site.
It appears that these elements could be binding sites for
the AGAMOUS transcription factor.
When we use another tool called Athena what we can do is,
we can look for the presence of known elements in
the promoters of our genes from a database called PLACE.
Question d asks us to comment on the number of elements in the promoter and
here we see lots and lots of elements, each one is depicted by a different colour.
However, only a few of them are actually significant.
And one of the significantly enriched
binding sites from the place database is in fact the AGAMOUS binding site.
And this is just based on a hyper- geometric test,
looking at the presence or absence of that binding site in the promoters of these AP3
coexpressed genes, relative to the overall count in all promoters.
So we would say that, again, this (AGAMOUS) could bind in that case, based on this output.
So question f asks: do you think that the two motifs from the Promomer analysis,
those two top 6-mers, are part of a larger motif?
So what we can do is you can take the consensus sites for
those two k-mers, motif1 and motif2,
and then use a tool called Cistome to view where those
motifs occur in the promoters of our AP3-coexpressed genes.
So we paste in our motifs here for those two 6-mers, we paste in the AGI IDs,
just the IDs, of the AP3-coexpressed promoters and
then we click Map, and this is the kind of output that we get.
We see that they're actually not that overlapping, so the green and
the red in this case denote the two separate motifs.
And we don't see a lot of overlap, and in fact we can ask what the overlap is here
by clicking on the merge button down in the bottom left.
And we actually see only a few cases where those motifs actually do overlap.
So they don't seem to be part of a larger motif.
Questions g and h deal with a database, a really nice open access
database called JASPAR developed by Wyeth Wasserman's group at
University of British Columbia, and others, and here we're searching,
because we saw that there could be an AGAMOUS binding site in the promoters of
those genes, we're asking whether or not, we want to
explore that information in more detail, the AGAMOUS binding sites.
So using JASPAR, which is nicer because instead of
containing just a consensus binding site for given transcription factor,
it actually contains a position specific scoring matrix.
So we know the preference at each position for binding.
We can use that information to understand our genes,
our coexpressed genes in a bit more detail.
So by specifying the species identifier 3702 in this search
interface here, you can actually pull up all the records for
Arabidopsis transcription factors that are in JASPAR.
And when we do that, we see the output here.
Here is AGAMOUS, the thing that we think is actually present in
the promoters of the AP3-coexpressed gene...
the cis-element that we think, or the binding site that we think is present in
the promoters of the AP3- coexpressed genes.
The identifiers you can just get from this column here, the first column.
And how was the specificity for the AGAMOUS entry determined?
So what we can do is we can click on that entry and under here, under the type,
we see that the method used to determine the bindings specify of AGAMOUS,
the AGAMOUS transcription factor was SELEX and we talked about that in the lecture.
All right, so now we revisit this question as to whether or
not AGAMOUS or AGAMOUS-Like 15 binding sites are statistically
enriched in the promoters of our AP3-coexpressed genes.
By taking that position-specific scoring matrix from
the JASPAR database, and using that to search, instead of
these consensus motifs which contain less information...
so the PSSM is a better way to search, and when we search with those,
we actually don't see any enrichment of those sites that would
match to the PSSM in the promoters of the coexpressed,
The AP3-coexpressed gene, so
we would probably say that AGAMOUS doesn't bind there, all right.
So next, we're using a tool called MEME to discover
motifs in a more probabilistic way. So Promomer is a word count method whereby
MEME is a more probabilistic method for discovering motifs.
And here, the output that we get describing these motifs is as
a PSSM, not just as a 6-mer word or something like that.
MEMEs stands for Multiple EM for
Motif Elicitation and what we're doing here in the case of MEME
is we're taking our set of AP3-coexpressed promoters, the actual sequences
we've provided, you're uploading those to the main interface and
then you are clicking Submit after adjusting a few parameters, as per the lab report.
And here's the output of the MEME algorithm, and
we see that there's a good motif.
I would call it a good motif because it's got a pretty good e-value.
A pretty low evalue in the promoters of the AP3-coexpressed genes.
In the second one, perhaps it's not as good,
but it's still below a 0.05 threshold.
Now does that motif match to anything in the databases
of known binding specificities for known transcription factors?
So if we use the CisBP database, and we take our
PSSM from this, for this, for this motif here,
we can query against all of the motifs that have been,
all of the binding specificities that have been entered into that database,
and see if there are any matches.
So, the top matches is actually to SOC1.
SOC1 one is known to be involved in regulating flowering.
AP3 is involved in flowering actually...
is involved in regulating floral development at least.
And if we look in the JASPAR database,
we actually see that FLC is one of the best matches (2017: so are AP1, AP3, PI and SOC1).
Perhaps not as convincing a match... I would say that
the SOC1 looks a little bit more like the motif that we discovered with MEME.
In this case, the FLC match,
you could argue that it's poly A rich whereas our motif is poly-dT rich.
But the reverse complement is poly A. It's also known obviously
that FLC is involved in regulating flowering in Arabidopsis.
So we could do some hand waving and say potentially either of those two
could be the targets or our discovered motif could be the targets of FLC or
SOC1 in regulating the AP3 coexpressed genes.
All right, now we turn to some human data.
And here we're looking at the promoters of insulin-coexpressed genes.
Obviously, insulin is important for regulating our metabolism
when we take in glucose, we get some insulin production going on.
This helps to regulate glucose levels in the blood.
And when we feed in the promoters of insulin-coexpressed
genes from human, we
get some pretty good looking motifs back again. So
at least this first one seems to be very quite believable
with a very low e value 1 times ten to the minus 76.
And that motif looks like, depending on when you run that
algorithm (there's an element of stocasticity),
when you run this algorithm you may get four or
five Cs followed by TGT and then followed by four or five Cs.
And the peculiar thing about that for this first motif
is that it occurs many, many times in the promoter of the insulin gene.
So this motif is represented by these red boxes
in this larger overview of where those motifs map.
And interestingly, if we look at what's known,
what the known motifs are present in the insulin gene,
we don't really see any sort of repetitive stretch from this particular publication.
So, here's our thousand base pair long
promoter and remember from the lab we said that it started 300 base
pairs upstream of transcription and continued 700 base pairs...
sorry, 300 base pairs down- stream of transcription and
continued 700 base pairs upstream.
Here's the corresponding map of known
motifs from this known publication by KN Daugherty and
in fact we don't see, as I mentioned, this repetitive element anywhere.
So we may have actually discovered a new element in promoter of the insulin gene,
so that would be cool.
Now, does that element actually match anything known in the databases?
So when we search using a tool called TomTom,
we don't really see any particularly great matches.
So this is the best match in
the JASPAR database these other Cs aren't here, present at the end.
In this case, the second best match.
Sure, there are a lot of Cs, but we're missing the Gs and
neighboring Ts on either side of that G.
So it could be a bonafide novel cis element in the promoter of insulin.
All right, by the end of Lab 6,
which comprised the labs, and the boxes, and the lectures.
You should know the main technologies for identifying a transcription factor
binding sites in vitro.
You should be able to name some methods for
identifying potential transcription factor binding sites in silico.
Such as word counting, these Gibbs sampling methods,
more probabilistic methods.
Should also understand how those word count and Gibbs sampling methods work.
And you should understand the differences between those two methods.
You should understand why we would want to generate results,
which referenced a background set of all promoters.
It's really nice to give you a feeling for how often or
how much over representation there is relative to background set.
You should also know how to use online tools, such as Athena or
JASPAR to identify known cis-elements in promoters of coexpressed genes.
And you should also be able to use the online version of MEME to generate
elements from a set of coexpressed genes promoters
