Assignment 13

Your name:
Your email address:

 

Assignments:

1. Using PRSS (or here) determine if there is significant similarity between proteins with the following accession numbers AAA23671.1 (D-Ala D-Ala ligase) and P04425.1 (Glutathione synthetase). Copy the fasta formated sequences into the PRSS window. Select 1000 shuffles, then click on "Shuffle Sequence" to start the comparison.

What is the E-Value (10000) of the comparison?

2. (10 minutes) Collaborate with your neighbor. One of you should do exercise #2, the other exercise #3. Share the results!

Do a PSI-BLAST search with the Glutathione synthetase (P04425.1 ) as a query (use uniProtKB/swiss-prot as database, and restrict the taxon to only search Bacteria (bacteria (taxid:2)) and select the PAM250 substitutions matrix). Under "Filters and Masking", turn on the filter for low complexity (tick/check the box), set the E-value cut-off for inclusion in the next round ("PSI-BLAST Threshold") to 0.00001 (or 1e-5) and change the maximum target sequences to 20000.
Note : Depending on the settings, PSI-Blast switches back and forth between the format and the result window. Place a checkmark in the box Show results in a new window (at the bottom of the page).

Note that the iteration number is listed at the top of the result page. One "launches" the next PSI-Blast iteration by looking for the "Run PSI-Blast iteration ? with max" at the top (or the bottom) of the list of significant matches, and clicking "Go".

After how many iterations (do not more than 5 iterations!) do you start to pick up D-Ala D-Ala ligase and Carbamoyl-phosphate synthase ?

Which other types of enzyme are included among the hits?


Note:Collaborate with your neighbor. One of you should do task #2, the other #3.


3. (10 minutes) Do a PSI-BLAST search with the D-Ala D-Ala ligase AAA23671.1 as a query (use uniProtKB/swiss-prot as database, and restrict the taxon to only search Bacteria (bacteria (taxid:2)) and select the PAM250 substitutions matrix). Under "Filters and Masking", turn on the filter for low complexity (tick/check the box), set the E-value cut-off for inclusion in the next round ("PSI-BLAST Threshold") to 0.0001 (or 1e-4) and change the maximum target sequences to 20000.
Note : Depending on the settings, PSI-Blast switches back and forth between the format and the result window. Place a checkmark in the box Show results in a new window (at the bottom of the page).

Note that the iteration number is listed at the top of the result page. One "launches" the next PSI-Blast iteration by looking for the "Run PSI-Blast iteration ? with max" at the top (or the bottom) of the list of significant matches, and clicking "Go".

After how many iterations (do not more than 5 iterations!) do you start to pick up carbamoyl phosphate synthetases, Glutathione synthetase, and Biotin carboxylase (aka Acetyl-CoA carboxylase A1)?

Which other types of enzyme are included among the hits?


What might be the reason for the different results obtained in tasks 2 and 3?


4. (30 Minutes) Do a PSI-BLAST (use uniProtKB/swiss-prot as database and restrict the taxon to only search archaea (archaea (taxid:2157))) search for 5 iterations, an E-value cut-off for inclusion in the next round ("PSI-BLAST Threshold") to 0.0001 with the following sequence (if this takes to long, use this PSSM linked below):

DO NOT use the gi number as query. The gi number refers to the whole protein, we want to use the intein sequence only!
>Pab_VMA intein from gi|7436316|pir||D75028
CVDGDTLVLTKEFGLIKIKDLYKILDGKGKKTVNGNEEWTELERPITLYGYKDGKIVEIKATHVYKGFS
AGMIEIRTRTGRKIKVTPIHKLFTGRVTKNGLEIREVMAKDLKKGDRIIVAKKIDGGERVKLNIRVEQKR
GKKIRIPDVLDEKLAEFLGYLIADGTLKPRTVAIYNNDESLLRRANELANELFNIEGKIVKGRTVKALLI
HSKALVEFFSKLGVPRNKKARTWKVPKELLISEPEVVKAFIKAYIMCDGYYDENKGEIEIVTASEEAAYG
FSYLLAKLGIYAIIREKIIGDKVYYRVVISGESNLEKLGIERVGRGYTSYDIVPVEVEELYNALGRPYAE
LKRAGIEIHNYLSGENMSYEMFRKFAKFVGMEEIAENHLTHVLFDEIVEIRYISEGQEVYDVTTETHNFIGG
NMPTLLHNT

DO NOT use the gi number as query. The gi number refers to the whole protein, we want to use the intein sequence only!


What types of enzymes do you get as hits?
Can you verify that the target proteins contain inteins? (How?)
Which E-value cut-off for inclusion in the next round did you choose?
What is the percent identity of the least significant hit added in the last iteration (clicking on the score in the table will jump to the alignment)?

Save the PSSM (Position Specific Scoring Matrix, or profile) from this search. To do that choose PSSM from menu inside the download link on top of the result page. Save the PSSM as an ASN file on your computer.
(If the iterations take too long, the PSSM after 6 iterations on the swissprot database is here, the PSSM after 5 iterations on the nr database is here*)

* I used the following command (commandline using the new blast+ package): psiblast -db nr -out out.inteinq -query inteinquery.txt -out_pssm inteinquery.pssm -out_ascii_pssm inteinquery.asci.pssm -inclusion_ethresh 0.0001 -num_iterations 5

Go to PSI-BLAST. Select BlastP (on top of the page). Paste intein sequence into query sequence box. Select the non redundant database.
Select the organisms to which the search should be restricted. You can select individual organisms or whole groups. (if you start typing, options will appear form which to select taxa)

Possible are

After selecting the genomes to search, go to Algorithm parameters and under PSI-blast options select and upload your PSSM from Question #7 .

As you start your search with a PSSM matrix already, you do not need to do iterations!

What are the results of your search? Did you get any significant matches? What are they?
If you have significant matches, does the match occur over the full lengths of both query and subject sequences? What is the percent identity?
Use Blink to investigate if the hits are indeed inteins.
What is your conclusion?
In your answer indicate - genomes searched, - number of significant matches found, - the E-values of these matches, and - the identity of these matches (i.e., are these probable inteins, or are they likely to be something else?).

5. Using PSSMs

SKIP THE COMMAND IN BLUE!!!!!

IS605.faa.txt is a FASTA formatted file containing an annotated IS605 transposase protein sequence from a Frankia genome.

We will use it to build a PSSM for this protein family, and then compare (mainly quantitatively) three searches:

  1. a blastp search of Frankia genomes for ORFs with significant matches to the sequence in the FASTA file,
  2. a PSI-blastp search of the collection of ORFs from the five genomes, and
  3. a PSI-tblastn search of Frankia genomes for significant matches to this PSSM

To do this, we will use the cluster. Establish a terminal connection to the cluster at via mcbsubmit.cam.uchc.edu (same username and password as previously) with the Bitvise SSH program installed on your computer. (Since the files are going TO the cluster in this case, we will use "curl" for convenience.)

A FASTA file with all the proteins from 5 Frankia genomes is here - fiveFrankia.faa.txt

A FASTA file of the nucleotide sequences of the 5 Frankia genomes is here - fiveFrankia.fna.txt

Note that these (currently 5) Frankia genomes were retreived from the Entrez ftp site. In this case, I retreived the *.faa (protein) files, and the *.fna (genome nucleotide) files.

Preparation: (bracketed parts are comments only!) 

	mkdir lab13								(make a new folder called "lab13")
	cd lab13								(equivalent of "double-clicking" a folder to descend "into" it)
	curl -O http://gogarten.uconn.edu/mcb3421_2018/labs/IS605.faa.txt		(curl stands for "see URL", this downloads the IS605 file to the cluster)
	                                                              	 	    (...you could also accomplish this with the Bitvise SFTP client)
	ls									    (make sure the file is there)
	cat IS605.faa.txt					    (...and that the contents look OK)
	srun --pty -p mcbstudent --qos=mcbstudent --mem=2G bash	 (we're about to do some serious computation, so onto a compute node we go...)
	module load blast/2.7.1                 (load the blast module)
	ls									    (once more to make sure the IS605 file is present in this folder)

To create the PSSM for IS605: (the psiblast command is long!...remember to scroll to the right to get all of it!)
SKIP THE COMMAND IN BLUE!!!!!

	export BLASTDB=/common/blast/data					(just to make sure BLAST knows where to find the "nr" database)
	export BLASTMAT=/opt/bio/ncbi/data					(...and the scoring matrices)
										(...ideally "BLASTDB" would be set in your .bash_profile, to avoid extra typing)
	psiblast -query IS605.faa.txt -db /isg/shared/databases/blast/nr -out is605.faa.blast -outfmt 6 -num_iterations 5 -out_pssm is605.faa.pssm -save_pssm_after_last_round
("-num_threads 2" says to use two threads -- similar to two processors/CPUs -- which should make it faster, at the expense of using twice the resources)

This takes quite a while, grab the PSSM from here - is605.faa.pssm:
curl -O http://gogarten.uconn.edu/mcb3421_2018/labs/is605.faa.pssm

Options for psiblast can be seen using 

	psiblast -help
or look at this psiblast-help.txt

In preparation for our blast searches on the Frankia genomes, we need to first create a searchable blast database: 

	curl -O http://gogarten.uconn.edu/mcb3421_2018/labs/fiveFrankia.faa.txt
  makeblastdb -in fiveFrankia.faa.txt -dbtype prot -parse_seqids
	curl -O http://gogarten.uconn.edu/mcb3421_2018/labs/fiveFrankia.fna.txt
	makeblastdb -in fiveFrankia.fna.txt -dbtype nucl -parse_seqids

To do a normal blastp search: 

	blastp -db fiveFrankia.faa.txt -query IS605.faa.txt -out blastp.out -evalue 1e-3 -outfmt 6 -num_threads 2
To do a PSI-blast search of the encoded proteins: 
	(get the PSSM first, because we skipped the command in blue, see above)
	psiblast -db fiveFrankia.faa.txt -in_pssm is605.faa.pssm -out PSIblastP.out -inclusion_ethresh 1e-3 -evalue 1e-3 -outfmt 6 -num_threads 2

As we are doing only the last iteration, an E-value cut-off of .001 is ok -- no danger of corrupting the PSSM.
Note, we need to add -evalue 1e-3, otherwise the e-values are reported in the table up to an e-value of 10!

You will receive a fancy warning message, but I believe it works just fine. More on the warning message is here.

To do a PSI-blast search of the 6 reading frames of the genomes: 

	tblastn -db fiveFrankia.fna.txt -in_pssm is605.faa.pssm -out psitblastn.out -evalue 1e-3 -outfmt 6 -num_threads 2

Counting the number of lines (corresponding to the number of significant matches; note, we had set the evalue to 10-5, and selected tabular output) in a file:

wc -l blastp.out
wc -l PSIblastP.out
wc -l psitblastn.out

or   wc -l psitblastn.out PSIblastP.out blastp.out

Also note (for fun purposes only) that the tblastn matches are not distributed evenly by genome: 

	cut -f2 psitblastn.out | uniq -c				(cuts out 2nd field, which is the target genome ID, and counts the matches for each of the 5 genomes)

How does the number of blastp matches compare to the number of PSI-blast and PSI-tblastn matches?

If there is a significant difference in the number of matches, can you think of a reason why this could happen?

BE SURE TO TYPE 

	logout								(log out from compute node and return to master node)
TO RELEASE THE QLOGIN JOB FROM THE QUEUE

 

    Finished?

    Type logout to release the compute node from the queue.
    If you you encountered problems in your session, check the queue for abandoned sessions using the command qstat. If there are abandoned sessions under your account, kill them by deleting them from the queue by typing qdel job-ID, e.g. "qdel 40000" would delete Job # 40000

 

    Send email to your instructor (and yourself) upon submit
    Send email to yourself only upon submit (as a backup)
    Show summary upon submit but do not send email to anyone.