OmicsConnect wiki

This page reports current issues and plans with omicsconnect, an extension on top of Observ. This work is in the context of EU-BioSHARE WP4.

NB this document formatted in trac wiki format, so we can include it online. Syntax:

• unsolved issues
• solved issues

  comments

showstoppers

• mrefs master/detail forms don't behave well. Examples:
  • adding new citations for study.
  • adding new measurement to protocol

discussion items

• Phenotype method <-> protocol / measurement
• Who does contribution: Person or an PersonAffiliation??

WormQTL datatypes review

v1 -> v2: notes

• analysis -> computeprotocol: needs merging work
• derivedtrait -> MISSING: observablefeature is what you want, but this needs to be typed in some way
• parameterset -> mref 'parameters' in computeprotocol: look into consequences not having sets for this
• binarydatamatrix -> MISSING: matrix backend specific, link to binary storage files
• gene -> MISSING: probably sequencecharacteristic? how is this typed? must be subclassed first? fields such as xref to Chromosome, bp location, etc via sequencerelation?
• parametervalue -> computeparameter: better than before with datatyping/freerange values
• chromosome -> chromosome: missing some handy fields such as 'orderNr' and 'isAutosomal' ('rank' in sequencerelation?)
• investigation -> study: OK
• probe -> [see: gene]
• data -> dataset: might need some extra fields such as ValueType?, ColumnType?, RowType?, Storage, etc
• marker -> marker: missing some fields: xref to Chromosome, cM location, reportsFor. 'seqlength' as int might be too small? (2x10^{9 when signed)}
• runtimeproperty -> MISSING: but make molgenis-wide build-in?
• dataname -> computeparameter: if values can be proper XREFs
• measurement -> [see: derivedtrait]
• sample -> MISSING: subclass of observationtarget?
• dataset -> [see: parameterset, if values can be proper XREFs]
• panel -> panel: OK
• snp -> [see: gene]
• datavalue -> computeparameter: if values can be proper XREFs
• parametername -> computeparameter: OK
• transcript -> [see: gene]
• investigationfile -> MISSING: but debatable use/implementation

wanted entities

Domain model:

observation

• Characteristic: 'is a' or 'has a' statements about an object
• ObservableFeature?: characteristic observed (is/has) or question asked (what is?)
  • Measurement: feature defined by description, unit, dataType, permittedValue
  • PermittedValue?: a categorical value state for Measurement
• ObservationTarget?: target of observation
  • Individual: ObservationTarget? consisting of one individual
  • Panel: ObservationTarget? consisting of group of individuals
• Protocol: procedure of observation on a defined set of features
  • discussion: many-to-many OR xref relation with Feature?
  • see PhenotypeMethod? discussion
• ProtocolApplication?: one bundle of values for exactly one target
  • xref to ObservationTarget? (e.g. Individual, but also SNP)
  • xref to Protocol
  • discussion: xref to DataSet??
• Observation: value relation between ProtocolApplication? and ObservableFeature?
  • xref to ProtocolApplication?
  • xref to ObservableFeature?
  • (no xref to Target, this is via ProtocolApplication?)

organization

• Study: a bundle of experiments typically published in a paper
  • discussion: bundle of Experiment, i.e. StudyElement??
• Experiment: a story of experimental objectives and outcomes
  • discussion: refers to dataSets?
• DataSet?: a group of ProtocolApplication?
  • discussion: homogeneous (i.e. using one protocol only?)
• SelectionCriteria?: conditions of relationship between Panel
  • xref to Source and Target panel
  • has: number of individuals, details
  • discussion: how to formalize

attribution

• Contact: an entity having contact details such as address
• Person: to refer to natural actors in the process
• Organization: organization actor
• PersonAffiliation?: association between individual and Organization
• Submission: the event of data contribution
  • dicussion: what are Submissable items? I.e. can have Contribution
• Contribution: metadata describing Role of Person in a Study (?)

references

• Hotlinks: hyperlinks to external? Or internal?
• Citation: literature reference
  • discussion: what entities can have references?
• OntologyReference?: reference to coding lists
  • Species: reference to species definitions
• AlternateIds?: alternative identifiers
• FeatureAnnotion?: qualitative association between Feature and OntologyTerm?
  • has 'exactmatch' and 'description'

genome

• SequenceCharacteristic?
  • Marker: genomic variable site
  • Allele: alternative alleles of marker
  • Genotype: valid genotypes for a marker
• SequenceRelation?

gwas central specific

• FrequencyCluster?
  • can be modelled as Protocol?
• GenotypeFrequency?
  • can be modelled as Protocol?
• AlleleFrequency?
  • can be modelled as Protocol?
• SamplePanel?
  • is a Panel with panelType=SamplePanel??
  • is a Panel with panelType=AssayedPanel??
• UsedMarkerset?
  • can be modelled as Protocol? or DataSet??
• EffectSize?
  • can be modelled as Protocol?

checklist of requirements

• study and person should have qualified 'contribution' denoting 'role' in study
• both study and data sets can be submissions (what else?)
• submission can have contributions (== role in submission)
• person can have more than one affiliation, how does that relate to role?
• remove fax and phone
• study design should come from an ontology
• need a way to link panels (e.g. sample panel, assayed panel)
• hotlinks to see only hotlinks related to a specific element, e.g only see experimental specific hotlinks in experiments?

FAQ

Q: Where do I add panel specific information e.g. population information, number of males/females, Geographical location

A: these are a ProtocolApplication? having panel as target

Q: where to add species?

A: species are special kind of ontology term