Working with SRS

Introduction to Sequence Retrieval System

Sequence Analysis Service, October, 12th-15th 1999

Introduction:

Information about molecular biology is available in a growing number of different biological databases. These different databases specialize in particular ways. For example all known enzymatic activities are contained in the ENZYME database, tertiary structures of known proteins are stored in the PDB databank, EMBL contains nucleotide sequences, SWISS-PROT contains protein sequences (mainly translated from the EMBL nucleotide database), the PROSITE database contains a dictionary of conserved sites and patterns in proteins, OMIM contains information on mutations involved in human genetic disease, etc.

Many of these databases have direct cross-references to other databases. For example EMBL entries are cross-linked to SWISS-PROT, PDB etc. SWISS-PROT has the most cross-links of all. These cross-references help to track down related information stored in another database. Of course, cross-indexing is only useful if you have a way to use the link - and this is where SRS comes in.

SRS makes a network of all the databases that have been made available inside it. It has a powerful indexing system to build indices, and a parser that can be adapted to parse all the very different formats used by the databases. (Unfortunately database developers have not bothered to use common formats). Once the index is built, the database is ready for probing with SRS using SRS's query language.

The query language supports the three boolean operators (BUTNOT !, AND &, OR |) and two link operators left link <, right link > . These operators allow complex queries and queries to be combined. The general syntax of the query language is

[database-field:keyword]
or

[database-field:keyword & keyword] > database

When you want to search with part of the keyword use the wild card '*' , example [swissprot-des:transmem*]

You can use SRS query language under the Unix command line interface (command getz ) or through a WWW interface (command wgetz ). Nearly everything in SRS can be done in either mode, but the ease and efficiency may vary.

We will concentrate on the WWW interface for most of our searches. However, with the command line interface ( getz ) we will see how to generate a small database where we want to get hold of a group of subsequences having some property in common.

The WWW interface has many different pages like

Select Library to select the databanks to search with
Query Form to fill out keywords and to select the fields to be displayed etc.,
Query Result to examine the result
Query Manager to manipulate the queries
View Manager to manipulate the way of displaying the result
LINK to link to other databases
Download Query to save the results to your disk
Databank to list all available databanks under SRS

and even a few more pages which will be explained during the practicals.

A typical database explained

SRS uses the flat file (text file) format for databases. A database contains one or more Entries. An Entry contains several different fields. For example a SWISS-PROT entry looks like the following (Some fields are omitted in this example).

Fig.1 Sample SWISS-PROT entry Click on the highlighted id name to see the full entry

_______=> Accession number | | ____=> name of the field (in this case identification) | | | | _____=> id name | | | | | | | ID ACH1_CAEEL STANDARD; PRT; 498 AA. --AC P48180; DT 01-FEB-1996 (REL. 33, CREATED) --- DE ACETYLCHOLINE RECEPTOR LIKE PROTEIN, ALPHA-TYPE PRECURSOR. | OS CAENORHABDITIS ELEGANS. | - FT TRANSMEM 231 252 POTENTIAL. | | SQ SEQUENCE 498 AA; 57169 MW; 570ECAC1 CRC32;
| | | MSVCTLLISC AILAAPTLGS LQERRLYEDL MRNYNNLERP VANHSEPVTV | | | DVDEKNQVVY VNAWLDYTWN DYNLVWDKAE YGNITDVRFP AGKIWKPDVL | | |__=> Sequence field | |__________________________=> Feature Table field | |____=> Description field

Getting Started

Login to any of your favorite machines.
Type netscape4. This brings up the netscape browser.
Open the following URL: You get this documentation online! http://www.embl-heidelberg.de/~seqanal/courses/predoc99/srstut.html
Click here (Click with the middle mouse button!) to start SRS with another window
You get the SRS home page. Now click on start button. You are in the Top Page where you can select databases of interest and continue.

Finding more about databases

Some people are scared when they first see this page! How about you! This is the main database selection page. Throughout the SRS session you will see the database names are 'hyper-lighted'. Clicking on them takes you to another page where you will see more information about the database. Also it shows you the different fields that this particular database has and the indexing date of the database etc., If you are lucky you will also get a WWW link to the 'home' site of the database.

Browsing the indices

Select SWISSPROT by clicking on the check box. Press the continue button.
Now you see the Query Form page. There are two types of query forms available. Click on the 'alternate queryform' button to see the other one. Choose the one that has all the fields 'hypertext link'. All the available data fields are highlighted and listed with the input field to be filled. Click on ID data field. You get the Data-field information page. It gives you information about the indexing date, status, number of id's etc., You can browse through this.
Type some id names (if you know any) or just click on List Values button. Check what you get.
Similarly try out other fields.

Exercise: Try to find out the other way of browsing indices by clicking on the database name.

Exercise: i) How many different types of molecules are present in the EMBL database ?
ii) How many molecule types are unannotated ?

Exercise: What are the different divisions in EMBL database ?

Querying the database

Go to Top Page. (Click on the TOP PAGE button)
Select SWISSPROT database by clicking the check box. Then press Continue button.
Type kinase in the All Text field and press Do Query button.
Check how many entries you get.
- Question: Is this going to be useful? If so when ?
Click on an entry name and check whether you get the full entry! Browse through this entry and see the cross-links to other databanks.
Go back to Query Form Page using Netscape's Back button. Type kinase in the Description field. Check the result. You can narrow down your search by querying the appropriate field. With the All Text field you get a lot of entries and with the description field you get a lot less entries.
- Why ?
Go back to Query Form Page. Click the Description field check box beneath Include in List. Press Do Query button.
You see that the Description field now appears on the query result! You can check the other fields like this. Try Keywords field.

Using the boolean operators ( BUTNOT, AND, OR ):

There are rather a lot of kinases with very diverse functions but you can narrow down your query using these operators.

Go back to Query Form page. Type tyrosine & kinase in the description field. Now you should get the total number of entries.
Go back to Query Form page. Type tyrosine & kinase ! receptor . And check the result!
- Question: did you manage to exclude all transmembrane receptor kinases?

Using the view

Go to Query Manager Page. You see all your queries are registered here. You can reinspect the queries.
Select the last query. Select the Sequence Simple view under the using the view option. Click on the VIEW button. Now you get the result in a table. You can add and remove the fields you want through the View Manager page.

Saving the query result

We will see how to save the query results to your disk. It is important to know how to do this. eg. if you want to make a multiple alignment of the sequences.

Reinspecting the query: Go to Query Manager page. select the query you wanted to save.
Change the using the view option to * Names only *
Click View button.
Now you will see the query result. Select one or more entries by clicking the check boxes.
Press save button.
Now you are in the Download query page.
Change use view option to * Complete entries * .
Change use mime type option to file.
Press Save Button.
Now the browser will ask you to enter a file name.
Enter some file name (eg., swiss.dat ) and press OK button.
The complete entries are stored in your disk. Check it!
Exercise: How do you save the whole list of your query result in a simple way ?

Linking the queries with other databases

SRS has two link operators. Namely '<' (left link) '>' (right link). The query expression

[swissprot-id:acha_human] > prosite
means link the swissprot acha_human entry to prosite! You get the prosite entry.

[swissprot-id:acha_human] < prosite
means the same above, but this time you get the swissprot entry. This makes sure that there is a cross-reference available for prosite.
Note: The PROSITE database entries each cover a whole protein family and all matching SWISS-PROT entries are cross-referenced.

Check the Library Network to see how different libraries are linked directly or indirectly! The numbers shown there are the number of steps needed to link them. You can click on the numbers to get a clear picture about how the intermediate libraries are used for indirect links between any two databases.

Go to the Query Manager page. Select the last query.
Click on the Link button.
Now you are in the Database Link Page . Select EMBL database and press the Continue button.
You get all the EMBL entries which are cross-linked from SWISSPROT.

Another simple way to link queries: In the Query Manager page, type q1 > embl and press Expression button. You can type any valid SRS query syntax here!

Exercise: Get all the kinase entries for which the 3D structure is known.
How many are there ? (PDB contains 3D information)
Exercise: Retrieve all transmembrane segments found in human proteins.

[swissprot-org:human] > [swissprot-FtKey:transmem]
Exercise: Retrieve all human proteins that have transmembrane segments.
[swissprot-org:human] < [swissprot-FtKey:transmem]

More Examples:

[swissprot-id:acha_human] > prosite > swissprot

This retrieves all the SWISS-PROT entries for the protein family to which acha_human belongs.!

enzyme < pdb

This very simple query gives all the enzyme database entries for which the 3D structure is known!

The concept of Sub-Entries and their usage

We now know how a SWISS-PROT entry looks like. Sub-entries are made from annotated subsequences! For example one of the Feature Table fields of SWISS-PROT entry 7UP1_DROME contains

FT   ZN_FING     236    260       C4-TYPE.

ZN_FING means Zinc finger. The corresponding subsequence (from 236 to 260) is

C R G S R N CP I D Q H H R NQ C Q Y C R L K K C

SRS treats this as a sub-entry. This is useful since you can collect and use them like any other database using other sequence analysis packages. (We will make a database of Sub-entries with the command line interface getz).

Exercise: Go to Query form page. get all the sub entries for zinc finger. Remember the FtKey field. Then click on one of the sub entries to see how it looks. Can you get the original (parent) entry?
Exercise: Retrieve all the intron subsequences from the EMBL database. How many entries are there ?

Searching other databases

We will try the OMIM database.
OMIM was originally a book (Mendelian Inheritance in Man) of short review articles on genes and genetically based traits of medical importance. Now it is on-line. Most of its information is text, so that relevant articles can be retrieved searching through keywords.

Searching OMIM can be fun!

Go to Top Page
Select the OMIM databank and press Continue.
To learn more about DMD (Duchenne Muscular Dystrophy) disease.
Type dmd in the Keyword field.
Have you ever wondered why do you sneeze ? Well, this time search with sneeze inside OMIM and find out about that. (You can even type 'ACHOO' )

Getting information from secondary sources

SRS can provide links to external web sites, such as Medline, or for example if a database is not supported at EMBL. Thus Swiss-Prot entries for yeast proteins have links to SGD at Stanford. Sometimes you can learn a lot by using these external links. We'll follow up links about the human inherited disease Phenylketonuria (PKA).

Go to Top Page
Select the Swiss-Prot database and press continue
Type in phenylketonuria.
Which of the listed entries has classical phenylketonuria ?
Use the OMIM and PAHdb links to find out about mutations causing PKA
- Questions
- Are PKA mutations dominant or recessive ?
- How many different classes of mutations (splice, missence, etc) can cause PKA ?
- Is it true that missense mutations are the rarest class ?
- Are the mutations evenly distributed through the sequence ?
- Do SWISSPROT Features list all classes of mutation ?
- For an even more striking example of human genetic disease, go to the Marfan Syndrome entry and repeat the exercise

getz: Command line interface to SRS

Login to any Unix machine, e.g. tau
type prepare srs
if prepare does not work, type source /app/prepare/prepare first and then type prepare srs
type ' getz -help'. If this does not work then you have problems! Report!

Try out the following commands! Remember you have to protect the query syntax with inverted commas ''.

 getz -libs 

   Lists all the databanks available under SRS.

 getz -info swissprot 

  Prints more information about swissprot library indexing data, 
  data fields etc.,

 getz '[swissprot-id:acha_human]' 

  Prints only the identification name. 

 getz '[swissprot-id:acha_human]' -e  

  Prints the entire entry. The default format is swissprot. 

 getz '[swissprot-id:acha_human]' -e -sf fasta 

  Prints the entire entry with fasta format.

 getz '[swissprot-id:acha_human]' -e -sf gcg 

  Prints the entry in GCG  format. 

 getz '[swissprot-id:acha_human]' -f 'id seq' -sf fasta 

  Prints only the id line and sequence with fasta format. 

 getz '[swissprot-id:acha_human]' -f 'id des seq' 

  Prints the Identification name 

 getz '[swissprot-id:acha_human] > prosite' -e 
 
  Prints the entire entry of prosite ( NEUROTR_ION_CHANNEL ).

 getz '[swissprot-id:acha_human] > prosite > swissprot'

Documentation for SRS query syntax available!
Creating databases
In this exercise let us try to create an intron database for viral sequences in the EMBL database. The main reason why we use command line interface for SRS is the network traffic on the WWW to make the new databank might be very large!
- Check what are the fields available to query EMBL database.
  getz -info embl
- The following command looks for the keyword intron in the embl viral subsection. Note: Beware the result might be very long, for safety, use a unix filter (less or more) wherever applicable
  getz '[embl-div:vrl] > [embl-ftkey:intron]' | less
  Can you count how many hits you got ?
- The above command lists only the entry names. We want the sub entries. So add some more fields "-f 'id acc ftkey seq' " etc.,
  getz '[embl-div:vrl] > [embl-ftkey:intron]' -f 'id acc ftkey seq' | less
- or simply add -e option!
- To store the result just pipe it to a file!
  getz '[embl-div:vrl] > [embl-ftkey:intron]' -e > intron.dat
- Exercise: Use the command line interface to repeat the searches you have done through the WWW.
Note: There might still be some bugs with getz, please inform me when you find them.
Online version of this material is available at:
http://www.embl-heidelberg.de/~seqanal/courses/predoc99/srstut.html
Chenna Ramu

Select Library	to select the databanks to search with
Query Form	to fill out keywords and to select the fields to be displayed etc.,
Query Result	to examine the result
Query Manager	to manipulate the queries
View Manager	to manipulate the way of displaying the result
LINK	to link to other databases
Download Query	to save the results to your disk
Databank	to list all available databanks under SRS