Patricia Chan gave a lecture on setting up the UCSC Genome Browser. genomebrowsersetup.pdf

• 32/64 bit Linux/Unix system
• CGI
• MySQL database
• Apache

Written in C and JavaScript (JQuery)

To install or mirror a genome browser on a new server: http://genomewiki.ucsc.edu/index.php/Browser_Installation

• MySQL Database
  • Each genome assembly has its own database
  • Most track data stored in MySQL

• /gbdb/<DB name>
  • Each genome assembly has its own local directory
  • sequences, wiggle track data, other large data sources (.bam files)

centraldb

• Contains all genome information.
• centraldb.dbDb stores the information
• centraldb.blatservers stores the info for two blat servers (one of DNA, one for protein sequences)
• Can be renamed in .hg.conf

Track Info

• Stored in the trackDb table in each genome's MySQL database.
• Based on trackDb.ra files as input source.
• The global trackDb.ra contains tracks that apply to multiple genomes.
• Similar to building a custom track on the genome browser.

Kent Code Base

• Required to set up the browser.
• Latest source code is in GIT repository.
• Need a CGI sandbox.
• make utils in ~/kent/src
• Binaries are installed in ~/bin/${MACHTYPE}

Browser Configuration File

• .hg.conf
• Contains MySQL user accounts and passwords, centraldb info, trackDb info.
• Required by Kent applications to connect to MySQL.

• Create /gbdb/newGenome directory for a new genome assembly
• Convert genome sequences from FASTA to 2bit format.
  • faToTwoBit chr1.fa [chr2.fa …] /gbdb/newGenome/newGenome.2bit
• FASTA input files must have UNIX LF character.

• create a MySQL database for the genome assembly:
  • hgsql “” -e “create database if not exists newGenome”
  • hgsql is a wrapper for passing SQL commands. The first argument is the database name.
• create a group table for the new database
  • cd ~/kent/src/hg/lib
  • hgsql newGenome < grp.sql
• create a chromInfo table
  • faSize -detailed chr1.fa [chr2.fa …] > chrominfo.tab
  • hgsql newGenome < ~/kent/src/hg/lib/ chromInfo.sql
  • hgsql newGenome -e 'load data local infile “chrominfo.tab” into table chromInfo;“
  • hgsql newGenome -e 'update chromInfo set fileName = ”/gbdb/newGenome/newGenome.2bit“

• Add an entry into the centraldb.dbDb table
• Add an entry into the centraldb.defaultDb table
  • Set the default assembly to use.
• Add an entry into the centraldb.genomeClade table
  • The clade the genome is associated with.
  • If the genome belongs to a clade that is not in the browser, add an entry to the centraldb.clade table.
• Add a description of the genome in an HTML file (/gbdb/newGenome/html/description.html)
  • Free formatted.

• Each genome database needs a trackDb table
• the global trackDb.ra is in ~/kent/src/hg/makeDb/trackDb
• genome specific trackDb.ra is stored in ~/kent/src/hg/makeDb/trackDb/<DB name>
  • can be stored in an alternate location.

• a hgFindSpec table is required for specifying search criteria.
• Search criteria for each track are also loaded from trackDb.ra.

• To run BLAT, gfServer for each genome must be started.
• Insert 2 records into centraldb.blatServers table.
• Make sure port numbers are different.
• If BLAT server is not run locally
  • rsync -v /gbdb/newGenome/newGenome.2bit blat_host:/gbdb/newGenome
• At the host machine, start BLAT server in the background.

• The previous steps are automated in the perl script make-browser.