Exercises on protein moltifs and domains

Go to parent Basic bioinformatics concepts, databases and tools#Exercises_during_the_training

Proteins exert their function through their specific 3D structure (tertiary and quaternary structure). Proteins that exert the same function often have similar (parts of) 3D structures since they're originally derived from the same ancestral protein. Similar parts of 3D structures correspond to a domain, which is a region in a protein that performs a specific function. Hence, when aligning proteins, stretches of similar AA residues (so called domains or motifs) can be detected. So, the first step to determine function of protein sequences is to look for known domains and motifs in the protein sequence that are related with a function.
Many webbased tools for this can be found on this page (and also on Expasy)

Searches for motifs or domains

Unknown motifs or domains

Exercise 1: HMMER for building HMMs from protein alignments

HMMER is a very user-friendly tool for generating and using hidden markov models. A HMM can capture nearly all information (residues and gaps) of a multiple sequence alignment. When searching a database with a HMM, we can reach greater sensitivity than by searching the database with a single sequence as BLAST does. Use hmmsearch to search a protein sequence database using an alignment or a HMM

It only accepts a few MSA fortmats: often you need to convert your alignment to the right format. You can use a multiple sequence alignment converter.

Note that HMMER contains four programs:

• phmmer: to search a protein database with your protein sequence
• hmmscan: to search a HMM that matches your protein sequence
• hmmsearch: to search a protein database with your HMM
• jackhmmer: iterative procedure to search a protein sequenc e database with a protein sequence or HMM. After first iteration, the hits are added to the alignment, the HMM is refined and the search is repeated...

The hmmsearch option is what we will use. You should have on your computer 12 histone H1 protein sequences from plants (from the UniProt exercises. In the previous exercise we have aligned these sequences.

Convert the alignment to a format that is accepted by HMMER.

Go to the multiple sequence alignment converter. The alignment is in CLUSTAL format and the format that is accepted by HMMER is Stockholm format. Click "clustal to stockholm" in the "Conversion map" section Open the CLUSTAL alignment in NotePad, copy the entire file (not only the alignment but include the title line that starts with "CLUSTAL") and paste it in the "clustal content" section Select "Protein" in the "Alphabet" section Click "Convert" The resulting stockholm file is displayed in the "stockholm converted content" section. Copy the converted alignment and paste it in NotePad Save as .sto file in NotePad

Now in hmmsearch select "Upload a File". Browse to the .sto file and click "Submit". Keep all default settings.



HMMER now creates a HMM of the alignment, and looks which protein sequences from UniProt (since we have selected to search against the UniProt database) match this HMM. Fairly quick, you get the results page showing the hits. Above you see a distribution of the E-values, and by hovering your mouse you can find more details about the contained hits. Such a profile is convenient to determine your threshold: if nice separate peaks exist, you can decide to set the threshold in the 'valley' between two peaks. Next in the results page a list of hits appears, with an E-value reported, similar as in BLAST.

How many hits are reported by this search?
1631 (as of June 2013)

Hence, this search is an alternative to a simple BLAST search. We have created a HMM from sequences from Swissprot (hence high quality sequences) and next we have used this HMM to very sensitively search the UniprotKB for protein sequences that show similarity to the conserved parts of the proteins in our alignment.
Small note: the PSI-BLAST does a similar thing, but with a PSSM (position-specific scoring matrix) instead of a HMM.

Exercise 2: Searching for sequences with low similarity using PSI-BLAST

Use one of the histone H1 protein sequences from the exercise above about HMMER to search with PSI-BLAST to find similar sequences in the RefSeq protein database. I used H1.1 from Arabidopsis thaliana. Go to the BLAST page at NCBI and search with PSI-BLAST against Viridiplantae RefSeq protein sequences. Please note that PSI-BLAST searches can take a lot of time !

What is the accession number of the best scoring new hit after iteration 2 ?

Fill in the BLAST form as follows: Click "BLAST". When the results page appears, it looks similar to a regular BLAST search, apart from a small box above the results list. This box is the power of PSI-BLAST: when you click 'GO' here, it will perform another BLAST search using a position-specific weight matrix constructed from the hits to score the similarities. You will see that iteration 2 will yield hits which are coloured in yellow ("below threshold in the previous iteration"): those hits were not found with the first search. You can continue to iterate this way, until no hits are labelled as 'new'. Hence PSI-BLAST can find more distantly related proteins. The accession number of the first new hit in iteration 2 is NP_001236157.

Training/tutorials on PSI-BLAST: http://sitemaker.umich.edu/microbial_genomics/files/psiblast3-3.doc

Known motifs and domains

Exercise 1: Ensembl

Information on the location of known motifs or domains in protein sequences can be found in many databases. One such example is Ensembl. Search the human F9 (Coagulation factor IX Precursor) gene and go to the transcript page of F9-001, its longest transcript (see exercises on Ensembl for details).

In which part (N-terminal or C-terminal half) of the protein, encoded by F9-001, does the peptidase activity reside ?
On the transcript page, go to the left menu and click "Domains & features". When you scroll down you see that there the domains responsible for the peptidase activity are located in the C-terminal part of the protein. The peptidase activity is responsible for the cleavage of factor X to its active form factor Xa.

A nice link to try out is the "Display all genes with this domain" link, e.g. try it for the Peptidase_S1 domain.



On the Transcript page you see now a graphical representation of all human chromosomes. On each chromosome, the positions of genes that encode a protein with a Peptidase_S1 domain is visualized by blue triangles. The location of F9 is shown as a red triangle.
Below this graphical representation, you see a list of all genes that encode a protein with a Peptidase_S1 domain. You can download this list by clicking on the Excel icon on the top of the list and selecting "Download whole table".

Exercise 2: NCBI's CCD

Search for protein AAF51293 on NCBI's Protein database.

What is the name of the domain in this protein ?
In the right Menu, in the "Analyze this sequence" section you can click "Identify Conserved Domains". The protein contains a SLC5 solute binding domain; hover your mouse over the domain to see more information.

Exercise 3: PROSITE characterization of human TPA

We have already analyzed the sequence of the human tissue type plasminogen activator (TPA) by downloading and aligning it in a dotplot. This protein has UniProt ID TPA_HUMAN and UniProt accession number P00750. This protein has well documented motifs, which we will look at using online tools
Of the many existing databases containing protein motifs and families PROSITE is the best curated and annotated. PROSITE allows you to:

• report the protein domains in your sequence
• report proteins containing a certain motif

We are going to analyze human TPA. In PROSITE you can provide an identifier instead of a sequence:

• Type 'tpa_human' in the bottom-left box under "Scan a sequence against PROSITE patterns and profiles"
• click "Scan"

Prosite will scan the sequence for all the patterns known in Prosite.



Results are returned in one long page. PROSITE contains motifs of two kinds: patterns and profiles.



Hence, in the output page you will find first the "hits by profiles" (or "matrices") (usually longer motifs) and then the "hits by patterns" (usually shorter motifs).

The PROSITE search engine seems to have done a fine job in finding all the domains. A summary figure on top provides a nice overview.

The page is extensively postprocessed and provides a lot of extra information beyond the mere matches.
In the description of the individual motifs you can hover your mouse over certain components like "DISULFID" (C-C bridges) to highlight them in green in the domain sequence

In the complete sequence, the C's are highlighted in green and the domain in yellow.

You can click on various links, e.g. click "PS50070"

to access the PROSITE documentation about kringles including a description of the domain and a list of proteins that contain the domain. From there you can click on "PS50070" (kringle domain) and "PS00021" (kringle motif) to access more detailed information, e.g. PSM of the domain, regular expression of the motif, list of UniProt sequences containing the domain/motif, list of sequences from PDB (for which the 3D structure was determined) containing the domain/motif.

Exercise 4: Retrieving all protein sequences containing a certain pattern

Prosite allows you to find all proteins that contain a certain pattern: the tool for this is called ScanProsite. The pattern must be written in a certain syntax, information on the syntax can be found here.

Download all SwissProt sequences with motif YRGS at the C terminus in FASTA format as 'scanprosite_yrgs.fasta'.

Go to ScanProsite Select the second option In Step 1: enter the pattern of the motif in the text box. With the help of the provided information (click on "Your own pattern" in the page or look at the slides), you know the pattern should look like this: Y-R-G-S> In Step 2: deselect "Include splice variants" At the bottom click "Start the scan" There are 23 SwissProt sequences containing this motif at their C-terminus. To download all matched sequences, go to the bottom of the Results page and click "Matched UniProtKB entries" In the screen that appears, all sequences but the isoforms are listed. Download them by clicking the orange download button and choose "Dowload" in the FASTA section.

Please note that you can also search in one specific protein sequence for a motif. In that case you select Option 3.

Does the tpa_human protein contain the YRGS motif at its C-terminus.
In Step 1: type "tpa_human" in the text box In Step 2: type "Y-R-G-S>" in the text box Click "Start the scan" You will see that TPA does not contain the motif at its C-terminus.

Let's get back to all protein sequences that do contain the C-terminal YRGS motif. You have downloaded them to a file 'scanprosite_yrgs.fasta'. Align the sequences.

Which two are 'aberrant' ?

Use Muscle to create a multiple sequence alignment of the sequences. View the results in Jalview. In the alignment, you should see the YRGS motif at the C-terminus. In Jalview you see that the two aberant sequences are Q9LVM5 (the third sequence in the alignment which does not align well at the start of the sequence) and Q5P1G0 (the first sequence in the alignment which does not align well at the end). These two sequences do not align well because they are much larger than the other sequences in our file.

Go back to the FASTA file, remove these two sequences (this is easily done in WordPad) and align again.
Search with the above created alignment for similar sequences in the RefSeq database using HMMER.

How many significant hits do you find ?

Download the alignment in FASTA format. Remember that you need to convert the alignment format: check format converter. Go to the HMMER website. Select "hmmsearch". This nice tool allows to input a alignment to search the RefSeq database at NCBI. But the alignment needs to be in Stockholm format. Select "RefSeq", upload the file in Stockholm format and click "Submit". More than 1500 hits are found! Check the sequence logo at the bottom to see if you can see the conserved YRGS at the C-terminus. Remember that you set a threshold on the E-value for significance and a threshold for reporting in HMMER. As a default: the threshold for reporting is higher than the threshold for significance, meaning that the tool will also report 'non-significant' hits. Below on the results page you can click "Jump to threshold page" if you want to see where the threshold was set. Only one page of the reported 17 pages of hits fall below significance threshold.

Exercise 5: search domains in human RIO2 via the InterPro database

InterPro is a composite database combining the information of all existing databases of protein families, motifs and domains. Some of these databases, like PROSITE, PRINTS and Pfam-A, contain motif and domain information. Other databases like ProDom and Pfam-B, contain information on protein families obtained via other methods (e.g. sequence clustering). The InterPro search engine searches the different member databases using their respective "native" search engines and then merges the results. Sometimes, Interpro finds motifs with different names from different databases that describe the same pattern: Interpro assembles them using an IPR-code!
Interpro works in the same two modes as Prosite does: you can search for a domain and retrieve all sequences that contain that domain or you can scan a sequence for an interpro domains (InterproScan).

Which domains do you find in human serine/threonine-protein kinase RIO2, with UniProt accession Q9BVS4 ?
Go to Interpro Enter the accession number in the search box and click "Search". Note the presence of many different Interpro signatures. Which domain is correct ? They all are!

Which database does domain "IPR018934" come from ?
Click the accession of the original database: "PF01163 (RIO1)" You are redirected to the Pfam database.

When you go back to the InterPro page, please note the "Similar proteins" link in the left menu



When you click this link, you are redirected to a list of proteins with similar domain architecture. You can download the sequences of the proteins of this list in FASTA format by clicking the "Export FASTA" button on the top right of the page.

Exercise 6: search domains in pig Q2L9D6 via Interpro

A very short protein from pigs has been found in blood (with UniProt accession Q2L9D6).

What is the Interpro accession of the family this blood localized protein of pig belongs to ?
The family this protein belongs to is IPR002449.

Which database does this family originally come from ?

You see that the family originally comes from the PANTHER database and that the retinol binding motifs come from the PRINTS database. However, when you click the Interpro accession to go to the details page, you see that Interpro also incorporates PIRSF info for this family. You see that one single Interpro family combines info from many databases. Hence, in articles preferentially provide Interpro accessions.

Can you tell me more about the species distribution of this family ?
On the details page click "Species" in the left menu. The family is found in Eukaryota only. There are 138 family members in Eukaryotes but in some Eukaryotes you have multiple members, like zebrafish and mouse.

Exercise 7: Search domains in Bacillus subtilis O32142 with InterPro

Retrieve the sequence of Bacillus subtilis (O32142). Search for domains/motifs using InterProScan.

How many domains/motifs does the sequence contain ?
You can use the Genbank accession to search Interpro. Type the accession number in the InterPro text search box and click "Search" The protein is linked to 2 families, 1 domain and 2 motifs

What is the relation between the two families that are returned by InterPro ?
Click on the InterPro ID of either of the two families to go to the InterPro record. There you can see that the second is a child of the first.

Which species contain the most members of the child family ?
Go to the InterPro record of the child family. In the left menu click "Species". There you see that there are three members of the family in Arabidopsis, maize and zebrafish.