Expression maps

Depends on VB deliverable(s):

• Expression data flat files
• Ensembl core database

Update triggered by external event(s):

Build instructions

At the moment this is Anopheles gambiae-centric We will build maps for other species when there are more experiments for them.

The Ensembl core database is used to extract "BioMart-like" information about InterPro domains and GO terms for each gene.

Making the map

The map is stored in MySQL (db=emap).

On bio-iisrv2:

Assuming that Expression data flat files is delivered, go to ~maccallr/vbge_preview/gextractor and gunzip the ...shifted-means.txt.gz file.

cd ~maccallr/expressionmaps/generation

The exact commandline can be found by working from the previous version's commandline in the database table emap.map

Run it with nice in a screen add quotes around map name - it takes several days (due to the 100 bootstrap maps).

The database connection config is in config.txt - you only need to change that if updating the GO database name.

GO & Interpro annotations

For new gene builds (or new GO/Interpro annotations) generate 'biomart.txt' from ensembl core database. It's not a real biomart file - it is just designed to look like one.

On bio-iisrv1 (or check out from SVN)

cd ~maccallr/vectorbase/expression/expressionmap/annotations
# edit ensembl.registry
./get_annots.pl > biomart-AgamP3.5.txt

copy file to ~maccallr/expressionmaps/go_ora on bio-iisrv2

Update GO database

Once every six months or as appropriate:

See http://archive.geneontology.org/termdb/README

Load into bio-iisrv2 mysql as database go_YYYY_MM_DD

Download http://archive.geneontology.org/termdb/latest/go_daily-termdb.rdf-xml.gz gunzip and rename to

~maccallr/ermineJ.data/go_YYYY-MM-DD-termdb.rdf-xml

GO over-representation analysis

cd ~maccallr/expressionmaps/go_ora

edit config.txt update godb and ermineJ_goterms db versions

biomart.txt must be the up to date version from the previous step

./make-ermine-psuedo-array-annotation.pl "VectorBase AgamP3.5 VB-2010-10" biomart-AgamP3.5.txt > AgamP3.5-VB-2010-10.ermineJ.txt

In a screen: ./run-go-ora.pl "VectorBase AgamP3.5 VB-2010-10" AgamP3.5-VB-2010-10.ermineJ.txt

Check that it's working. Results will go into the DB. You can start setting up the webapp while it's running. It takes a few hours.

Backups

mysqldump the emap database to the ND-rsynced directory on bio-iisrv1

Setting up the webapp

cd to apache htdocs/ExpressionMap/Anopheles_gambiae

sudo svn checkout svn+ssh://base.vectorbase.org/usr/local/svn_repository/expressionmap/prototype BETA-VB-2010-10

edit the config make sure biomart.txt is the correct version

On release day

Update the symlink htdocs/ExpressionMap/Anopheles_gambiae/current to point to new version