A program to make InterMine easier to install.

• user is computational biologist, NOT naive user and NOT expert user
• 27 May to mid-July

General Outline

1. run make_mine script
2. read project XML
3. add/configure existing sources
4. configure xxmine.properties for DBs and tomcat
5. run project_build script
6. indicate where data can be downloaded (this isn't essential)
7. add new sources
8. basic web configuration (if time)

model browser - similar to MISO

• reads in model from core.xml, etc and additions files included in project.xml
• tree on left
• fields listed on top right, each tagged with source name (not visible yet)
• graph of model
• JTGraph uses BSD license - ok?

installing InterMine

Simple Java installer wizard.

• ignore all sysadmin issues
  • assume tomcat, postgres installed, running correctly
• add/update MINE.properties in .intermine
  • prompt to create the database, give psql command
  • ignore web properties for now, leave defaults
• MakeMine
• add/edit BioSources
  • read project.xml
  • add/remove projects from (predetermined list), set organisms and data location
    • organisms to load: D. melanogaster, C. elegans
    • data location: /data/biogrid
  • assume data available and valid, use default config for priorities and primary keys
• generates project.xml, view model
• run project_build script

adding your own data to intermine

Add new source - See make_source

• prompts for converter type, eg obo
  • has user written converter yet? NO
• prompts the user for type of each data, eg "Gene"
  1. checks existing model then SO for data type
  2. displays
     • what is already in the model (model browser?), if anything
     • need all fields for keys and priorities
     • message about potential conflicts
       • use org.intermine.metadata.Model and and org.intermine.metadata.ClassDescriptor?
     • enforce Java naming conventions, eg. featureType not feature_type
  3. validates reverse relationships
  4. generate example model to see what it would look like
     • inheritance
     • suggest better data type? eg. "Fields in this object match Transcript ..."
  5. generates additions file
• autogenerate keys
  • eg "Which identifier fields are always unique?"
• autogenerate priorities
  • eg "UniProt creates proteins too. If they differ, [] use UniProt [] use my source"
• add to project.xml
• run project_build script - Can't do this until they write the converter