[MUSIC]
This exercise is going to be on the
PsiBlast tool at NCBI.
PsiBlast stands for position specific
iterated
blast, which is a specific form of
blast, designed to find distant homologs
of a particular protein or DNA sequence.
Our objectives for this exercise include
setting up the parameters of PsiBlast,
specifically to get the right threshold
for future iterations of PsiBlast.
We will enter a protein sequence to run
the first
iteration of PsiBlast, and then from those
results, we will
be able to compile a group of sequences to
run
a second iteration to find other members
of that protein family.
Finally, we will be looking in more detail
how to
interpret those results, which can be a
little bit tricky.
First, we will retrieve the protein
sequence that we were going
to use for PsiBlast, and that sequence is
available at NCBI.
Specifically, we'll be using a transposase
from Pyrococcus furiosus.
Pyrococcus furiosus is a hyperthermophile
that grows at 100 degrees Celsius.
It comes from a volcanic vent.
The organism was first isolated off the
coast of
Italy from a volcanic vent on the
Mediterranean floor.
Now I'm going to click FASTA to get the
actual sequence, and here is the sequence.
What you might notice is that the first
line is descriptive, describes the
transposes for Pyrococcus furiosus.
And then below that are a series of
letters,
which represent the amino acids in that
particular protein.
Okay.
You might notice that the top line, which
I have highlighted, is the description.
The description says that this is a
transposase from Pyrococcus furiosus, and
then below that
are a series of letters, and these letters
represent the amino acids in a particular
sequence.
Now the the FASTA formatted sequence
includes both the amino acids
sequence and the descriptions, so I'm
going to copy and paste that.
And now I'm going to go and do the
PsiBlast search.
To do that, I've got to go to the blast
page at NCBI,
and you can Google that or type in NCBI,
for National
Center for Biotechnology Information,
.NLM, for
National Library of Medicine, .NLH, for
National
Institutes of Health, .gov, and then slash
last.
That will take me to the main blast page,
and I'm going to click protein blast.
And if you notice, by protein blast,
you'll see a series of algorithms.
The algorithms include, blast p, psi
blast, phi blast, and delta blast.
So, we're going to click protein blast,
and the first thing I'm
going to do is paste in the sequence into
the window.
So there is my, the transposee is from
pyrococcus furiosus.
Okay.
Now I'm going to select PsiBlast.
[BLANK_AUDIO]
And I'm going to choose my database.
My database, in this case, is going to be
Reference Proteins.
It's a more high quality database than the
NR database.
The NR database stands for non-redundant,
but there are plenty of redundant
sequences.
It's a huge database with a lot of low
quality sequences.
We'll switch to the reference proteins,
and one thing I want to do
before actually running the PsiBlast is
you'll
notice algorithm parameters in the
lower-left corner.
We're going to expand that window, and I'm
going to change the threshold.
Scroll all the way down.
You'll see the default PsiBlast threshold
is 0.005.
I want to be a little more stringent than
that.
I'm going to use 1 times 10 to the minus
4th to make sure
that I include only members of the
transposes family when I do the second
iteration.
So I'm going to change that to 0.0001.
Okay, so I have my protein sequence.
I have my database.
I selected PsiBlast, and I've changed my
threshold.
One thing I think I want to do is limit to
a particular type of organism, so the goal
of our exercise might be to find homologs
or
relatives of this protein in a particular
type of bacteria.
So I'm going to enter Lacto Bacillus,
which will give me lactic acid bacteria.
Many different types, and you see that
that's
taxonomic ID 1578, so I'm going to click
that.
And now the results should come up
shortly.
[BLANK_AUDIO]
Okay, so here are the results, and you
notice
that there are a series of alignments
better than
threshold, and there are a whole bunch of
relatives
of the pirachocas transposes in
Lactobacillus.
And then you also see a series of proteins
that scored worse than the threshold.
Now typically for blast, if a, if an e
value, is less than 1 times
10 to the minus 4th, such as this one, 6
times 10 to the negative 5.
We can consider that a homolog or a
relative of the protein that we entered.
So everything above this line we can very,
very
confidently say, yes, these are homologs
of that pyrococcus transposase.
Anything below the line we're not quite as
sure of.
[BLANK_AUDIO]
Okay, but we see that there are some
transposases there.
So what we can do is examine the results.
Notice that almost all of these results
are from transposases.
Even some of the ones that are worse than
threshold are transposases, but
notice that everything above the line,
every protein above the line, has a check.
On the right side in that check box, and
if you notice the header under
worse for threshold says select for side
list, anything below that threshold is not
checked.
So what PsiBlast is about to do in the
second iteration is develop a
position specific scoring matrix or possum
as we commonly call them.
The possum is a mathematical
representation of the sequence that we
started with and all of these relatives
that are above threshhold,
and if I wanted to manually exclude a
protein, I say,
oh, yeah, maybe that's not a relative, I
could do that.
I could also take a sequence that's worse
than threshold.
I'll actually take the second one here.
It's a transposase.
Its e value is 1 times 10 to the minus 4,
probably right on the border.
So maybe I'll include that in the
PsiBlast, and now when I hit go, I'm going
to run a second version of blast, a
second iteration, except not with a query
sequence.
Now with this possum, that mathematically
represents all the
of the results in the first iteration of
blasts.
Okay, and these results should come
shortly.
Okay, now what we notice is, first of all,
a bunch of very, very strong e values.
2 times 10 to the minus 122.
I'll talk more about that in a second.
You'll also notice that some protein
sequences are
highlighted in yellow, and the sequences
that are
in white, without the yellow highlighting,
are protein
sequences that were found in the first
iteration.
Any new sequence that was only found
in the second iteration is highlighted in
yellow.
And we see some new sequences with very,
very strong e values, and I want to
caution you
about that because the nature of the
second iteration's
very different from the nature of the
first iteration.
In the first iteration, we started with
the, transposase sequence for pyrococcus.
And, when I ran that first iteration, I
said that anything with an e
value of less than 1 times 10 to the minus
4 is a homolog.
So, now I see e values much, much less
than 1 times
10 to the minus 4, but let's remember, in
the second iteration.
We're not running the query sequence from
pyrococcus.
We're now running a possum that represents
both
the pyrococcus sequence and a bunch of
other transposes.
So these e values say that the possum was
matched very, very
well, which suggests that it's probably a
member of that protein family.
However, in some cases, a possum might not
quite accurately represent
a protein family, maybe because a
nonrelative got into that mix,
and it's usually best to treat these new
yellow sequences as punitive
homologs, meaning that they should be
further checked for homology.
And you notice, in this case, that almost
all of the sequences highlighted in yellow
say transposase, which tells me that they
are probably indeed members of that
protein family.
So we can see that PsiBlast is a powerful
tool for finding potential homologs of
protein or DNA sequences that might not be
found by the first iteration of blast.
Blast is not 100% sensitive, so it will
not
necessarily find all homologs in an
initial database search.
Psi-blast is a powerful way to identify
potential homologs that may
have been missed by the first round of a
blast search.
This concludes our PsiBlast tutorial.
For more information, please contact us
in the Center for Biolotechnology
Information.
Thank you.
[MUSIC]