Outcomes of MAGE best practice dicussion Based on notes taken Wed. Dec 3rd at Jamboree, based on notes made by Paul Spellman Experiment Package FactorValue does not have a way to specify which channel has which factor values. How do we solve this? TIGR -- Joe White, use channel info as NVT Angel -- Require separation of PhysicalBioAssay by channel We expect a synthesis by Friday Afternoon Recommendation to apply FactorValues to all BioAssays (Physical,Measured and Derived) Null factor values -- All bioassays should have a value for each factor Associations Experiment to BioAssay: All Bioassays should be associated to the experiment ExperimentDesign to BioAssay (TopLevelBioassays): Should be Derived unless only measured unless only physical Experiment to BioAssayData; All BioAssayData should be associate to the experiment BioMaterial Package BioSource DO NOT have Treatments Treatments are CREATING events Treatments should have a parent BioMaterial (No clear use cases where it shouldn't) BioSamples can have new characteristics if necessary -- USE CASE mutagenesis which causes change in genotype LabeledExtracts can be created by treatments of LabeledExtracts USE CASE Example, make labeled mRNA, separate by size, run on arrays Treatment vs ParameterValues -- Major focus of MAGE2, nothing to be done now 1 ActionMeasurement per Treatment -- use the same action with the same order more than once, or use protocolApplication BioAssayPackage Structure Users should use the Hybrization (BioAssayCreation) Event BioAssayTreatment BioAssay to BioAssayTreatment should be 0..1 BioAssayTreatment should point to a new BioAssay, including ImageAcquisition Events (but may be forgiven) BioAssayMap For MIAME purposes must be included Channel Use the Dye name (so it can be related to Compound) PhysicalBioAssay can have channel MeasuredBioAssay and DerivedBioAssay associations to channel are discouraged MeasuredBioAssay and DerivedBioAssay to Data should be strongly encouraged Tentative endorsement for Angel's solution for ability to use ExperimentFactor for distinguishing which ExperimenFactors are associated with each channel BioAssayDataPackage DesignElementDimension -- reuse when possible DesignElementDimension Descriptions -- Nothing necessary, other ways to get this information Reference Channel -- USE ExperimentFactor with values of Reference or Experimental SummaryStatistics USE CASE How many features were saturated in the red channel... MGED Ontology will provide a set of names and types to use ArrayDesignPackage Reporters whose exact sequence is unknown: e.g. an oligo sequence that is based on a Genbank Acc No, should NOT have ImmobilizedCharacteristics --RULE for an ArrayDesign there is only one FeatureReproterMap per Reporter --RULE for an ArrayDesign there is only one ReporterCompositeMap per CompositeSequence RULE the number of Maps should be as small as possible How do we use Primers -- Create as BioSequences throughImmobilizedCharacteristics One of FeatureGroups, ReporterGroups and CompositeGroups should havespecies For ArrayExpress All DesignElements must belong to a Group and allGroups must be linked to an ArrayDesign ***Action Item*** For confidence indicators e.g. allow 58/60 base matches for oligo to gene Use the map Use the MGED Ontology for DesignElement ControlType All Features for a FeatureReporterMap must belong to a single ArrayDesign Individual Feature Failures should use ArrayManufacture rather than generate a new ArrayDesign BioSequence_ref pointers: Allow the use of LSID to determine BioSequences. There are a limited number of sources to be determined ArrayDesigns_ref pointers: Allow the use of LSID to determine ArrayDesigns. There are a limited number of sources to be determined ArrayBatch information should be recorded in the ArrayManufacture Name Other Back end cardinalities are unreliable e.g. OntologyEntry to itself, Node to itself. Identifier case sensitivity -- Must be enforced, all Identifiers are case SENSITIVE Only same source submissions are likely to reuse Identifiers