MAGE and MAD or MAD about MAGE

1
MAGE and MAD(or MAD about MAGE)
Joseph White(jwhite_at_tigr.org)The Institute for
Genomic Research www.tigr.orgMGED 6September
3-8, 2003
2
MAD and MAGE

• Database Design Issues for MAD
• Integrating MAGE into MAD
• MAGE Adapters MAGE-Perl
• TIGR_Madam MAGE-Java

3
MicroArray Database Schema
4
Conceptual Schema MAD
5
TIGR Microarray Databases
12/11/02
6
Design Considerations

• Normalized database tables
• Clones are related to Wells which are related to
  Spots on a Slide (Array)
• Experiment 1 Slide 2 Probes
• Expression table is huge must re-index
  frequently
• MIAME compatability extend existing tables, and
  add tables where necessary

7
Signals, Spots and Clones
Normalized tables
Slide map
Analysis
8
Clones, Genes and Oligos
9
MIAME Relational Schema
10
MAD vs. MIAME

• MIAME was designed as the model of public
  databases for microarray data.
• MIAME does not capture certain critical
  information e.g.
• Clone set information
• Plate-well to slide row-column data
• Clone and gene annotation information
• MAD lacks certain information as well e.g.
• MAD lacks contact and publication information.
• MAD treats an experiment as a single
  hybridization.
• MAD does not adequately treat quality control
  data.
• We do not store images in the MAD database
  itself, but store file names and locations.

11
Experiment and Hybridization Parameters
12
Probe Data
13
Studies, Experiments, and Replicates
14
Studies, Experiments and Replicates
15
MAGE OM for Array Data

• Purpose Provide and object model to capture data
  for a MIAME data repository.
• Accept data in XML format from any database, i.e.
  schema independent.
• Load data into MIAME database.
• Export data from MIAME database in XML format.

MIAME
16
MAGE Adapters
TIGR_MAGE.pm
17
TIGR_MAGE.pm

• COMPOUND_CY3 / COMPOUND_CY5
• CHANNEL_CY3 / CHANNEL_CY5
• QUANTITATION_TYPE_DIMENSION
• CONTACT hash
• make_identifier()
• make_person()
• pake_protocol()
• get_channel_from_filter()
• convert_date()
• get_measurement()

18
make_identifier

• sub make_identifier
• my(type,args) _at__
• my auth argsauth 'tigr.org'
• croak "make_identifier must specify authority"
  if ! auth
• croak "make_identifier must specify namespace"
  if ! argsns
• if(exists TYPEStype )
• croak "make_identifier must specify type
  name" if ! argsname
• return "LSIDMAGEauthargsnstypeargsna
  me"
• elsif (type eq 'Zone' )
• croak "make_identifier must specify type
  name" if ! argsname
• croak "make_identifier must specify type row"
  if ! argsrow
• croak "make_identifier must specify type col"
  if ! argscol
• return "LSIDMAGEauthargsnsargsnamet
  ypeargsrowargscol"
• elsif (type eq 'LabeledExtract')
• croak "make_identifier must specify
  LabeledExtract name" if ! argsname
• croak "make_identifier must specify
  LabeledExtract label" if ! argslabel
• return "LSIDMAGEauthargsnstypeargsna
  meargslabel"

19
make_identifier

• sub make_identifier
• my(type,args) _at__
• my auth argsauth 'tigr.org'
• croak "make_identifier must specify authority"
  if ! auth
• croak "make_identifier must specify namespace"
  if ! argsns
• if(exists TYPEStype )
• croak "make_identifier must specify type
  name" if ! argsname
• return "LSIDMAGEauthargsnstypeargsna
  me"
• elsif (type eq 'Zone' )
• croak "make_identifier must specify type
  name" if ! argsname
• croak "make_identifier must specify type row"
  if ! argsrow
• croak "make_identifier must specify type col"
  if ! argscol
• return "LSIDMAGEauthargsnsargsnamet
  ypeargsrowargscol"

20
make_identifier needs

• my _at_tmp_types qw(Experiment BioAssay Array
  BioMaterial ArrayDesign
• ArrayManufacture PhysicalArrayDesign
  Hybridization Protocol
• BioEvent Image FeatureDimension BioAssayData
  Security
• SecurityGroup BioAssayDimension Feature MAGE
  BioSequence
• Reporter CompositeSequence ReporterCompositeMap
  FeatureReporterMap
• ReporterGroup CompositeGroup FeatureGroup
  Hardware MeasuredBioAssayData
• MeasuredBioAssay ExperimentDesign
  ExperimentalFactor
• )
• foreach my type (_at_tmp_types) TYPEStype
  ''

21
make_identifier

• elsif (type eq 'LabeledExtract')
• croak "make_identifier must specify
  LabeledExtract name" if ! argsname
• croak "make_identifier must specify
  LabeledExtract label" if ! argslabel
• return "LSIDMAGEauthargsnstypeargsna
  meargslabel"
• elsif (type eq 'Treatment')
• croak "make_identifier must specify Treatment
  source" if ! argssource
• croak "make_identifier must specify Treatment
  target" if ! argstarget
• return "LSIDMAGEauthargsnstypeargsso
  urceargstarget"
• else
• croak "make_identifier unsupported type
  type"

22
makeOE

• sub makeOE
• my (category,value,uri,db) _at__
• my oe BioMAGEDescriptionOntologyE
  ntry-gtnew()
• if(! category ! value) return oe
  
• oe-gtsetCategory(category)
• oe-gtsetValue(value)
• my accn substr(uri, index(uri,"")
  1 )
• accn value if ! accn
• if(uri db)
• oe-gtsetOntologyReference(
• BioMAGEDescriptionDa
  tabaseEntry-gtnew( accessiongtaccn,
• 
  databasegtBioMAGEDescriptionDatabase-gtnew(
• 
  identifiergtmake_identifier('Database',
  authgturi, nsgtdb),
• 
  namegt"Databasedb",
• 
  URIgturi, )
• ) )
• return oe

23
MADAM

• Java v.1.3.1 Application
• MySQL v.3.23.49
• InnoDB
• Schema provided
• Released Sept, 02
• www.tigr.org

• Application will run on a laptop with a P3 650
  MHz processor and 256 Mb RAM.
• Requires 0.5 2.0 Gb of disk space for InnoDB
  tables.
• Runs on most flavors of Windows (not tested in
  Win 95).
• Installer will do all necessary work including
  creating DB.

24
(No Transcript)
25
ACKNOWLEDGEMENTS
John Quackenbush Principle Investigator Alex
Saeed Madam, MultiExperiment-Viewer Vasily
Sharov Madam, Spotfinder Jerry Li Madam,
Barcode reader Joseph White Madam, Array
Databases Tracey Currier Experiment Designer,
Web MEV Wei Liang Miner, MIDAS John
Braisted Web MEV Nirmal Bhagabhati MEV
Statistical Analysis Mathangi Thiagarajan User
Support Eleanor Howe User Support Special
Thanks Jason Stewart Perl Guru