Notes on Biological Databases - Fall 2008 | CMSC 423, Study notes of Computer Science

Download Notes on Biological Databases - Fall 2008 | CMSC 423 and more Study notes Computer Science in PDF only on Docsity!

CMSC423: Bioinformatic Algorithms,

Databases and Tools

Lecture 5

Biological databases

Biological databases

What data gets stored?

• DNA

• string of letters
• quality information, maybe chromatograms
• location of genes (ranges along a chromosome)

• Proteins

• string of letters
• protein domains
• 3D coordinates of each atom

• Pathways

• graph of interactions between genes

For all – often store link to scientific articles related to data

How the data get accessed

• Gene by gene/object by object – targeted at manual

inspection of data

• usually lots of clicking involved
• simple search capability
• similarity searches in addition to text queries

• Bulk – targeted at computational analyses

• often programmatic access through web server
• most frequently – just bulk download (ftp)

EMBL European Molecular Biology Lab.

• European version of NCBI

• BioMart query builder

http://www.ebi.ac.uk/embl/

Expasy proteomics server

• Home of Swisprot and other useful information on proteins

http://www.expasy.org

Genome browsers

• UCSC Genome Browser – http://genome.ucsc.edu

• ENSEMBL Genome Browser – http://www.ensemble.org

• Gbrowse http://www.gmod.org

Direct database access - SQL

• CHADO schema – www.gmod.org

Programmatic database access

use DBI; my $dbh = DBI->connect("dbi:Sybase:server=SERV;packetSize=8092", "anonymous", "anonymous"); if (! defined $dbh) { die ("Cannot connect to server\n"); } my $mysqlqry = ; $dbh->do("set textsize 65535"); my $qh = $dbh->prepare($mysqlqry) || die ("Cannot prepare\n"); $qh->execute() || die ("Cannot execute\n"); while (my @row = $qh->fetchrow()){ processrow($row); }

NCBI programmatic access

• http://eutils.ncbi.nlm.nih.gov/entrez/query/static/eutils_help.html
  • must write your own HTTP client (LWP Perl module helps)
  • queries go directly to web server
  • data returned in XML
• http://www.ncbi.nlm.nih.gov/Traces/trace.cgi?

cmd=show&f=doc&m=obtain&s=stips

• stub script provided (query_tracedb)
• queries still go through web server
• data returned in a variety of user selected formats
• For both, limits are set on the amount of data retrieved, e.g. less than

40,000 records at a time

• Download procedure:
  • figure out # of records to be retrieved ("count" query)
  • read data in allowable chunks
  • combine the chunks