BNFO 301 
Introduction to Bioinformatics
Genome Analysis
Spring 2014 


Search for FMRP in Drosophila

A couple of weeks ago you had the idea of studying Fragile X Syndrome in Drosophila. The plan was to use the known human FMRP protein and then use it to scan the Drosophila genome to find a similar gene. Then it's just a matter of mutating the gene in Drosophila, and investigating what effect such a mutation may have on fly mental function.

Then you got stuck. No Drosophila genome means nothing to scan. So for the past two weeks you have been sequencing the Drosophila genome. Now that's behind you, and you have a genome sequence in hand. The time has come to continue with the plan.

1.   From the course web site, click on Resources and Links and then on NCBI (National Center for Biotechnology Information, a repository of vast amounts of information).

2.   Click on the down arrow to expand the choices for the Search box from All Databases, and select Protein.

3.   Enter FMRP into the white box, and press Search (or press Enter). 

4.   You might expect to pull up an entry for the human FMRP gene, but no such luck. The search returned a few 100 entries (!), some from flies – which is interesting (You'll probably have to scroll down to see them). Let's try to reduce the number to something more manageable.

5.   Click the Advanced Search tab. Type FMRP into a white box next to All fields. To limit the search to the human genome, go to the second All fields box, click the down arrow to expand the choices from All fields, and select Organism, then type human into the box to the right. Finally, choose AND (it's probably already specified) to specify that you're looking for entries that contain both the word FMRP AND the organism human, click Search.

6.   I got about 60 entries now. Certainly an improvement, but surely humans don't have 60 different FMRP proteins! Part of the problem is that we're searching multiple overlapping databases and getting the same protein back multiple times. To cure that, go to the list of Source Databases on the left side of the screen and click a single database (I chose GenBank).

7.   Now I'm down to 16 entries, but most don't look very interesting. The remaining are different from each other in ways that need not concern us now. Click the entry with the accession number AAH86957.

8.   Lots of information here. One important item (scroll down a bit) is a reference to a journal article that describes the work that led to the results you're looking at. Clicking on the PubMed number leads you to an abstract and a link to the full length article -- potentially very useful! Remember how to do this! But what we're after right now is the protein sequence, seen at the bottom of the page. The sequence is given according to the one-letter amino acid codes (see course web site, Links and Resources, Genetic Code, for a list of them).

SQ1. Is this the right sequence? Scrolling back to the top of the page you'll see something labeled 'ORGANISM' and a list of terms starting with "Eukaryota". I'm willing to admit I'm a eukaryote. Chordata, Craniata, Vertebrata... yes, I have a backbone... but am I also a Haplorrhini and a Catarrhini? What do they mean?

SQ2. Under FEATURES (scroll down), there are three Regions given. What might they mean? Write down the three region names.

SQ3. What amino acid begins the protein (scroll down to the sequence)? Why?

9.   The display is good for some purposes -- the numbering makes it easy for humans to find what amino acid is at a specific positions -- but computers prefer straight sequence, without numbers or spaces. To get this, scroll back up to the top of the page. You'll see near the top a menu under the Display Settings button, and there you'll find GenPept checked. The format you're looking at is called GenPept (for "GENbank PEPTids"). Change the format to FastA by clicking he down arrow to open a menu and then clicking the button next to FastA and then clicking Apply.

10.   You may already know what FastA format is, but in any event, if you've gotten this far, you're looking at it: One line of documentation preceded by ">" and multiple lines of sequence. If you want to save this file, don't use the browser SAVE function, or you'll get a mess of html browser goo. Instead, click Send to, then click the File radio button. arrow, and finally click Create File to save the file to an appropriate directory on your computer. For now, just copy the sequence, with or without the documentation line. (In passing, you could also have obtained a savable file by clicking FastA (text) in the previous step)

11. Now we're ready to do the search. You can initiate the search using the sequence your looking at immediately by clicking Run BLAST in the right column, but for reasons that will become apparent, take a longer route to the same end. Open up the NCBI icon at the top of the screen in a new tab (on a PC, right click the icon) to return to the NCBI home page, and click BLAST in the Popular Resources box at the right of the page. BLAST (Basic Local Alignment Search Tool) is undoubtedly the most widely used bioinformatic tool in existence. We'll talk about what it does and how it does it, but for now, let's use it to find the fruit-fly gene we want. We have a protein sequence, and we want to find a fruit-fly protein, so look under Basic BLAST and click protein blast.

12. Paste the FMRP sequence into the Search box, where you're invited to enter a FASTA sequence (or upload the file, if you saved it) (or type in the accession number), and click the Blast button near the bottom of the page. You'll probably get a page pretty quickly showing conserved domains in the protein -- take a look at them and reconsider SQ2... More on this another day. For now, wait a bit more to get the full search results. These results may take tens of seconds to come up or more or less, depending on the time of day and alignment of the stars.

SQ4. (While you're waiting...) What do you think Blast doing?

13. Again, way too many hits! Each red line in the Graphic Summary Section is a very good hit.

SQ5. About those thick red lines... What happens when you mouse over one of them?

Each line in the Descriptions section gives a link to one of those hits. We'll cut them down in a moment, but first,, scroll down until you reach the Alignments section, which should look something like:

This shows the alignment of the protein we submitted (human FMR1 protein), i.e., the Query, to the protein that Blast found, the Subject. The letters to the right look like an amino acid sequence (check the query sequence -- it should be identical to the sequence you copied). The sequence to the right of "Sbjct" is the amino acid sequence of a similar protein found by Blast. The line in between show all the amino acids in common. Looks like the similarity is pretty good! Indeed, if you look at the Identities statistic above the sequence, you'll find that the sequences are 100% identical. The best hit happens to be to the FMR1 protein of Homo sapiens. This should come as no surprise. The protein most similar to FMR1 is FMR1 itself!
 
Continue to scroll down and you'll see similar proteins from other organisms: maybe chimpanzee (Pan troglodytes), gibbon (Nomascus leucogenys) dog (Canis familiaris), pig (Sus scrofa), cow (Bos taurus), orangutan (Pongo pygmaeus), mouse, rat,... the amazing thing is that this protein is nearly identical amongst mammals. The same is true of most protein. We share a common toolbox.

SQ6. Take a look at the best match in Pan troglodytes (chimpanzee), which will be XP_003317789. How many differences are there between the human and chimp proteins? (Examine the statistics at the top of the match) Where is the difference and what is the amino acid change? You'll have to peer very closely at the middle line, between the query sequence and subject sequence, for a position that does not contain a letter.
SQ7. Take a look at the best match in Myotis lucifugus (what is that?). How many conservative changes are there? (compare the number of positives to the number of identities). These will be represented by '+' in the middle line. How many nonconservative changes are there? These will be represented by ' ' in the middle line. How do conservative and nonconservative changes differ from one another? Why are these specific amino acid alignments considered conservative or nonconservative? Any other changes besides these?
SQ8. What sense can you make of the organisms associated with the hits? Why these and not others?

14. Unfortunately, none of the matches shown come from flies. Let's help the process. Click Edit and Resubmit at the top of the page to go back to the Blast submission page (where you pasted in the protein sequence). You'll see at in the middle of the page a section called Choose Search Set. In the box marked organism, start typing Drosophila, and click Drosophila melanogaster. Then click the Blast button again.

15. In several seconds... much better! Now you have only a few hits, (scroll down) all from Drosophila (of course). Here's the top hit:

SQ9. How good is this hit? What's the percent identical amino acids with respect to the human FMR1 protein you used as the query? Does this number correspond to your visual perception of the number of identical amino acids?
SQ10. There doesn't seem to be nearly as many identical amino acids as in the previous case. How likely do you think it is that such a match might have arisen by chance? Could this be some random protein that happens to be a little bit similar?

There are several entries that have the identical sequence, from three different databases (emb=European Molecular Biology Laboratory; ref=RefSeq; and gb=GenBank). You'll notice that the degree of similarity between the human and fly FMRP protein is much less than between human and mammalian proteins. Only 45% of the amino acids are identical, according to the statistics above the alignment. Is this degree of significant? The Expect value (close to 3e-115... this changes over time) gives you an indication. It says that if you search a database with the size and amino acid composition similar to the one you actually searched but with the sequences randomized, then a match this good or better would arise with a probability of 3 in 10^115 (1 with 115 zeros after it!). In other words, it couldn't possibly have arisen purely by chance in the lifetime of the universe. We'll return later to Expect values, since it is very important to understand what they mean and what they don't mean. By the way, since the GenBank database grows with time, the E-value also changes, so it may not be the number I gave.

16. OK, we've got the protein. But that's not good enough. You can't clone a protein. We need the gene to clone, mutate, and then put it into a fly to make a mutant fly that we can test for physiological function. Click the GenPept link, right above the statistics for the match. This should get you to the protein sequence (verify this by scrolling down to the bottom of the page -- do you see an amino acid sequence or a nucleotide sequence?). To find the gene sequence, look at the DBSOURCE (database source) field near the top of the page. A protein sequence is almost never obtained by direct sequencing of the protein but rather by computer translation of the nucleotide sequence. The DBSOURCE is given as NM_169324.2. Click that link.

17. This brings you to a page called "Drosophila melanogaster FMR1... mRNA".

SQ11. Check. Does this page give you the sequence of RNA?

The source of the protein sequence was virtual translation of a cloned piece of mRNA (since you can't clone RNA directly, it was first made into a DNA copy, or cDNA).

Scroll down to the field called CDS (CoDing Sequence -- don't ask me why anyone would think this is a proper abbreviation!). It says that the coding sequence (some would say gene) extends from position 569 to position 2614 in the cDNA sequence.

SQ12. Test that. Scroll down to the cDNA sequence (under ORIGIN) and find position 569. Is that position plausibly the beginning of the gene? How about position 2614. Is that position plausibly the end of the gene?

18. This may seem a roundabout path to get to the gene. Why not start with the human gene (instead of the human protein) and use it to pick out the fly gene? OK, we'll do it that way. Go back to the NCBI home page. Make sure Nucleotide is specified in the box at the top of the page. Enter FMRP as the search term, click Search, and, as you did in Steps 3 through 5, confine the search to humans and to GenBank and limit the search to GenBank.

19. Choose the first complete sequence that encodes FMRP (recalling what "FMRP" stands for) -- it will say "complete cds" in the descriptor line -- and is long enough to contain the sequence you found earlier. Avoid FRMP interacting protein, which is different from FMRP itself. Get the sequence from it as before.

SQ13. Why does the sequence end with so many A's? (perhaps the annotation can provide some insight -- take a look at the misc_difference fields)

20. Find your way to Blast, repeating Step 11, but this time choose nucleotide blast, using a DNA query to search a database of DNA sequences. Again paste the sequence into the window. (This time you may or may not need to change the defaults, specifying that the database is not the human genome but rather Others). Under Program Selection, Optimize for, choose "Somewhat similar sequences". Now click the Blast button. When the results are returned (maybe a minute), you'll see that there are tons of hits as before, but this time there are many mammalian sequences with multiple differences with respect to the human sequence. But never mind that -- we're after flies. So go back to the submission page and specify Drosophila melanogaster and Blast again.

SQ14. How long is the best hit?

SQ15. How good (E-value) is the best hit? What does that mean?

Why such a difference between the protein and nucleotide blasts?

SQ16. How can it be that the protein search gave long matches with extremely low expect values (definitely not merely by chance) but the nucleotide search found just tiny snippets?

One last trick. Get back to the BLAST front page, either starting over at NCBI or using the browser back button several times. This time choose BLASTX, which takes your nucleotide sequence and translates in all six possible reading frames before comparing the translated sequences against all known proteins. Paste in the human FMRP nucleotide sequence, choose Drosophila melanogaster as the target organism and go. (This might take about six times as long as your previous Blast)

SQ17. You used as query the same sequence as a moment ago, used as target the same organism... why are the results so different?

SQ18. Suppose that you have a mysterious phage gene and you'd like to know what it might do. Your plan is to use the gene to search for similar genes amongst all that are known. What general advice can you now offer yourself about the best way to achieve your goal?