Documents
Resources
Learning Center
Upload
Plans & pricing Sign in
Sign Out

Teleconference

VIEWS: 12 PAGES: 18

									Information Representation Working Group Face-to-Face
Meeting

October 18 and 19, 2010
Leader: Lisa Schick
Facilitator: Robert Freimuth

Executive Summary
The IRWG F2F Meeting was held to continue to work through the LS DAM
harmonization with BRIDG 3.0.2. Classes, attributes, associations and cardinality
were all assessed. New development to handle issues raised were generally
considered out of scope. Additions and changes to accommodate the Generic Assay
Model were reviewed. The intent of the model to be generic enough to handle all
assays means that in some cases the LS DAM class will be the parent of a BRIDG
class and there will be requests for some changes made to BRIDG SCC. Also, the
group concluded that subdomain views of the model were premature and not
needed.

Action Items
DAY 1
Action Items*:
Post Meeting or Long Term
     Need a business rule to describe how to use
       DocumentumentVersionRepresentation URL.
     Have Dave Hau at an IRWG meeting DONE
     What is the scope of data?
     Refer to FuGE model for Data, ExperimentalFactor, ExperimentalStudy
     Revisit Experiment attributes post planned and performed
    
For Day 2
     PointofContact – there are 2 approaches on the table. DONE
     Structural role to associate DocAuthor to person DONE
     Factor association to Experiment or through ExperimentalItem DONE
     ExperimentalItem cardinalities on the other end DONE

To   do on the Model from Day 1
    Research specific locators for Place
    Place - keep identifier (DSET) and add name(ST)
    Place - Make location recursive
    Protocol - Remove CategoryCode and subcategoryCode and add typeCode
    Material
     • Remove nanoSizedIndicator
     • Rename materialType to typeCode
     • Remove actualIndicator
      Keep effectiveDateRange
     • Revise definition of functionTypeCode
                 “The characteristic behavior of a material that results from the
                    chemical and physical composition and properties of the entity”
    • Add examples: kinase, RNAse
    • BRIDG SCC add Material.fxn.TypeCode
•   Product
       • Ask SCC to clarify class definition of Product with respect to FDA list
       • Remove nameCodeModifiedText
       • Remove pre1938Indicator
       • Remove classCode
       • expirationDate Inherit changes in Material
•   Equipment
     remove “age” in favor of an “activeDateRange”<IVL:TS>
     remove “location”.
     Follow up Place and Material association. Place should be associated at
       Material level.
     Remove availableForEval, OTC, returnedToReprocessorDate
    • Make validationDate a DSET
•   Data
    • Add creationDate (TS)
•   ExperimentalFactor
    • Look at FuGE to model ExperimentalFactor value
•   ExperimentalStudy
    • Keep typeCode(CD) and add designType (SC) and use DSET
    • include systems like disease, pathway in definition
    • change purposeStatement to description
    • Look at FuGE for ExperimentalStudy
•   Experiment
    • Keep designType
    • Use name, not title
•   ExperimentalItem
    • Associate to Protocol
    • Use just Quantity:PQ **DAY2 conclusion, not DAM level
    • Remove name:ST (have typeCode)
•   Software
    • Add licenseEffectiveDateRange (TS.DATE.FULL) accepted, check data type IVL
       TS
    • need to create association between software and document.
    • Add name:ST
    • Add licenseType and remove openSourceFlag (BL):
    • Not sure what the conclusion was on expirationDate (TS.DATE.FULL): The
       date (and time), assigned by the manufacturer, on which the product should
       not be used.
    • Add Datatype = IVL<TS.DATE>
    • Add
       o version(ST)
       o buildNumber (ST)
       o buildDate (TS.DATE )
       o licenseTypeCode (CD)
       o licenseEffectiveDateRange (IVL<TS.DATE>
       o licenseKey (ST)
       o Content(ED)
•   PointofContact
    • Use the defintion - Person or a department serving as the coordinator or focal
       point of an activity or program
    • Use attribute typeCode [vs. roleType]
•   Associate DocumentumentAuthor to Person
•   Associate Protocol and DocumentVersion directly
•   Removed subclasses of ProtocolDocument and ProtocolDocumentVersion
•   ExperimentalFactor and ExperimentalStudy – cardinality 0 to many
•   Association should be 0 to many from experimentalItem to animal, but 1 to many
    going in the other direction
•   Add association of Identified BiologicEntityGroup and ExperimentalItem
•   Associate ExperimentalFactor to ExperimentalStudy and Experiment.
•   Add association of Identified BiologicEntityGroup and ExperimentalItem
•   Associate ExperimentalFactor to ExperimentalStudy and Experiment.
•   Move attributes of qualify person, remove qualifiedperson

Day 2
 Experimental Item definitions revisit since they are not considering role class.
   Cardinality will need to be checked. (Role Material, Software, Animal)
 ExperimentalItem -update definition of typeCode:CD to: A code representing the
   role an entity plays in an experiment.
 Caught typos in definitions when reviewing Joyce’s comments
 Take to BRIDG SCC to make their study as a specialization of our
   ExperimentalStudy
 typeCodes not bound
 Add examples to show the LS ExperimentalStudy is the parent class to BRIDG
   class
 Make the intention of the generic assay model explicit. The intention of the
   model is to be so generic that any type of experiment or assay is captured –
   beyond omics.
 Justify subdomains by capturing issues with entity naming especially as we go
   into new activities.
 Take a recommendation to BRIDG SCC to change the datatype of
   DocumentVerisonBibliographicDesignation from ST to ED. By doing this, if you
   specify the MIME type, you can specify the format. In ED you can also choose to
   use string (it will be value).
 Add locatorValue:ST and LocatorType:CD to Place
 XP to handle locator value and type in Place
 Add association of Place to Specimen
 Add association of Place to Container
 Add Container class as a Product and associated to Place
 Lisa to work with Mukesh and Jenny on container.
 Lisa to review action items and bring any issues to the group on Thursday


Conclusion**:
    Associate DocumentumentAuthor to Person
      Associate Protocol and DocumentVersion directly
      Leave DocumentVersion and associations the way they are until we can work
       through repercussions of rolling some of the classes up.
      Removed subclasses of ProtocolDocument and ProtocolDocumentVersion
      Leave BibliographicDesignation as is. Potentially revisit ST or ED ** DAY 2
       Conclusion to use ED
      ExperimentalFactor and ExperimentalStudy – cardinality 0 to many
      Submission date and release date not needed for Experiment
      Association should be 0 to many from experimentalItem to animal, but 1 to
       many going in the other direction
      Add association of Identified BiologicEntityGroup and ExperimentalItem
      Associate ExperimentalFactor to ExperimentalStudy and Experiment.
      Leave Protocol and Experiment relationships as they are. Consider flavors of
       activities downstream, especially definedActivity.
      Move attributes of qualify person, remove qualifiedperson
      POC as a class accepted and the Organizational Contact attributes should be
       added to it. Include the constraints.
      Keep DocumentAuthor to Person association, but note there is no connection
       of a person to organization. See parking lot.
      ExperimentalItem - quantity is not at the DAM level.
      Have link from ExperimentalItem to Protocol and Experiment. Material list for
       protocols can be captured through an experiment template – see experiment
       typeCode.
      There is an association between Factor and Experiment and between Factor
       and Study. Leave as is.
      Not enough views yet for subdomains.
      Use ED for BiobliographicDesignation
** Some of things labeled as conclusions actually require action and appear in both
lists

PARKING LOT:
    DocumentVersion association with finding, protocol and data (and
      specializations of document and DocumentVersion)
    Roll up document id into document
    Relationships of Person, Organization, Point of Contact DONE
    Specializations of DocumentVersion
    Potentially revisit ST or ED for BibliographicDesignation DONE
    Relate experimentalFactors to Parameters
    Associations between Experiment and Activity and also between Protocol and
      Activity subclasses
    Associations between ExperimentalStudy and Activity and also between
      ExperimentalStudy and Protocol
    Issue in consistency of handling date in the model
    Related to Material - develop a physical characteristics model similar to
      caNanolab that covers at least mass, volume, solubility etc
    Generic id
    In Data, createdBy is data provenance
    POC affiliation (where is this person coming from).
       Affiliate (structural) may apply to DocumentAuthor



Meeting Notes

DAY1

Review of the model:
Document
Class is from BRIDG
Q. What does the content attribute hold? How does this work in a database?
A. It is for something pushed through a wire, you may not do it this way, but it is a
way to store a document electronically. Conceptually it keeps the content for the
document.

Discussion:
It seems odd to have encapsulated data and have the URL in the
DocumentVersionRepresentation class. It seems circular.

Q. Could you have the content or not?
A. Yes, leaves it optional to have the content or reference it. ED already has a URL in
it, so it can be used to store the URL only. ED doesn’t support a document stored in
multiple places. This is due to the NCI localization. (Heading towards using all of
them, but have to operate on what we know).

AI: need a business rule to describe how to use
DocumentumentVersionRepresentation URL.


Place
Class is from BRIDG
Use Case is MSDS – you need a physical copy on hand.

Q. Should physical be a qualifier to place?
A. Jenny prefers location

Discussion of URL being a place, vs. where a piece of equipment is physically
located.
PhysicalLocation is used in NucelicAcidSequence, so this is why we went with Place.
DocumentAuthor
From BRIDG without attributes. LS added attributes but it is not yet associated to
person.
It is a Functional role (no author without a document).

Q. Should this be associated with Person or do we want a structural role?
A. A use case is a PI running a study – you would want to be able to make the link
from the PI to the documents by that PI.

Discussion:
BRIDG has structural role of research staff, also have health related staff. It gives
the developer a way to restrict the list to choose from.
Research Staff: [read definition]. The designation as research staff would have to
occur before someone gets the ability to be an author. We would need to tweak the
definition.
Attributes typeCode, postal and telecom address (in your organization)
Structural role would also give you an associated with your organization

Q. What if you have multiple structural roles? How does this come together for one
functional role (author)?
There will be many documents authored by people who are not research staff
(brochure, MSDS, etc.).

Conclusion: Associate DocumentumentAuthor to Person

Q. How do you know what organization the person is associated with?
A. You can infer in the model via associations. Suggest an affiliation role (different
than member) because you need to associate via a structural role.
Parking Lot


Data and Finding and protocol to Document version association
Q. are these valid?
Discussion
ProtocolDocumentumentVersion – what is specific to it?

Conclusion:
Associate Protocol and DocumentVersion directly
Leave documentVersion associations


DocumentVersion
SpecimenCollectionGroup and DocumentVersion (i.e. diff path reports or informed
consent)
Q. Why do we associate this with DocumentVersion?
Discussion –
Finding to Document vs DocumentVersion– save all 10 copies you saved as you go
along.
Suggesting indirect vs. direct?
Document saying SNP not known to be causal, but this changes.
Do we need the Document class as an aggregator? It has typeCode which doesn’t
change.
Document and DocumentVersion are in BRIDG
Roll document id into document case against it is ii and primary indicator
Conclusion: Leave DocumentVersion and associations the way they are until we can
work through repercussions of rolling some of the classes up.

PARKING LOT:
DocumentVersion association with finding protocol and data
Roll document id into document

AI: Have Dave Hau at a meeting


Subclasses ProtocolDocument and ProtocolDocumentVersion
ProtocolDocumentVersion has no attributes, consider deleting.
Could be many other subclasses if we follow this course.

Conclusion: Removed subclasses of ProtocolDocument and ProtocolDocumentVersion
PARKING LOT: Specializations of DocumentVersion


BibliographicDesignation
This has string in BRIDG
Is the content a reference to a publication
Is there structure or business rule?
Add journal article subclass?
For SCC - ED might be a good data type for XML vs ST

Conclusion: Leave BibliographicDesignation as is. Potentially revisit ST or ED




Generic Assay
Experiment
ExperimentalStudy
ExperimentalItems
ExperimentalFactors
Software
Data
Material
Equipment
Protocol association
POC


ExperimentalStudy reviewed
Association to Experiment, ExperimentalItem, Protocol, ExperimentalFactor and POC

Experiment
Associations to Data, ExperimentalStudy, ExperimentalItem, Protocol,
ExperimentalFactor and POC
Synonyms: assay, procedure
Q. do you have to put in both dates in the range?
A. No, ongoing has not stop

Conclusions:
ExperimentalFactor and ExperimentalStudy – cardinality 0 to many
Submission date and release data not needed


ExperimentalItem
Associations to:
Animal – should be 0 to many from experimentalItem to animal, but 1 to many going
in the other direction
Discussion of cohorts. Have BiologicEntityGroup to cover it.
Each piece of equipment is and Experimental Item
Create instance id for biologicEntityGroup without requiring an individual animal.
Associate biologicEntityGroup and biologicEntity relationship and change to zero to
many.

Conclusions:
Add association of Identified BiologicEntityGroup and ExperimentalItem




ExperimentalFactor
Associate ExperimentalFactor to experimentalStudy and Experiment.
ExperimentalFactorValue class – do we want to move into ExperimentalFactor as a
set? Or do we need to associate it to anything else in the model, like data?

FuGE separates them out and associates to data
It will cover individual measurements
Factor is the agent – it effects the measurements results. FactorValue is the
condition.
Discussion of the class – model explicitly rather than elsewhere in the model
Leave it open to doing multiple assays vs. 1 experiment.

Q. Item is agent, want to capture factor and values – scope with agent. Other times
a factor is not scoped to an item.
Discussion - associate to experimentalFactorValue to data – this links to how the
data is captured vs. experimentalItem and experimentalFactorValue. ISA-TAB
discussed.
Disputing factor direct association to experiment. Look at what others have done.

PARKING LOT
Relate experimentalFactors to Parameters (already on the parking lot list)
Factor association to Experiment or through ExperimentalItem

Q. Can the same data serve in different experiments?
A. Yes. E.g. image


Protocol
Activity - Specialization of DefinedActivity?
Reviewed BRIDG and it is a set of templates
Discussion of the need for DefinedActivity based on PlannedActivity ( Perform based
on planned)
Value as touch point to BRIDG, need to link for caNanoLab
Defined activity means you are talking about protocol. Consider protocol is a
specialization of activity
Concern that you limit to type of activity as you go through.
Experiment performed and planned activity and Protocol from defined activity
Should it be Protocol as activity, Experiment to activity?

Conclusion: Leave as is. Consider flavors of activities downstream, especially
defined.
AI: Associations between Experiment and Activity and also between Protocol and
Activity subclasses
AI: Associations between ExperimentalStudy and Activity and also between
ExperimentalStudy and Protocol


Point of Contact
This is a Functional Role
Useful for software
Associated with qualified person from BRIDG
Discussion of Organization class
Specimen use case for knowing who gave you that sample
Q. How is it different than organizational contact?
A. It is from the organization Point of View
Q. Is there an association between organizational contact and PointOfContact?
Looked at BRIDG Performer
Q. Does this have to be a specific person?
A. No, help desk.

Three ways to capture pointofContact
       Organization POC
       Person not in organization
       Entity that is not a person but is a POC

Discussion: Person to POC, organization to POC, Move attributes of qualified person,
remove qualifiedperson. PointOfContact can be Person or organization

Conclusion/Options:
Person to POC, insert person contact (specialization). 0 to many
organization to POC, insert Organization contact between the two (specialization). 0
to many
Move attributes of qualify person, remove qualifiedperson
Or
POC can be Person or organization
Constraint so it is organization or person.
Person and Organization connected by OrganizationContact (to be renamed)

Attribute Review – Action Items (See Associated PowerPoint
Deck for Notes)
Place
   AI: research specific locators for Place
       Can a place be used to generate a summary of all items inside of? (ie all
    associations to place should be bi-directional) TBD
   keep identifier (DSET) and add name(ST)
              Make location recursive

Protocol
   Protocol renaming can occur in the layered view
   Remove CategoryCode and subcategoryCode and add typeCode


Material
•   Remove nanoSizedIndicator
•   Rename materialType to typeCode
•   Remove actualIndicator
•   Keep effectiveDateRange
•   Parking Lot: Issue in consistency of handling date.
•   Revise definition of functionTypeCode
        • “The characteristic behavior of a material that results from the chemical
           and physical composition and properties of the entity”
        • Add examples: kinase, RNAse
•   AI: BRIDG SCC add Material.fxn.TypeCode
•   Parking Lot: develop a physical characteristics model similar to caNanolab that
    covers at least mass, volume, solubility etc
Product
•   AI: Ask SCC to clarify class definition with respect to FDA list
•   Parking Lot – generic id
•   expirationDate Inherit changes in Material when we do those.
•   Remove nameCodeModifiedText
•   Remove pre1938Indicator
•   Remove classCode


Equipment
   remove “age” in favor of an “activeDateRange”<IVL:TS>
   remove “location”.
   AI: follow up Place and Material association. Place should be associated at
    Material level.
   Remove availableForEval, OTC, returnedToReprocessorDate
   Make validationDate a DSET


Data
   Add creationDate (TS)
   Parking Lot: createdBy is data provenance
   AI: What is the scope of data?
   AI: refer to FuGE model

Experimental Factor
   AI: to look at FuGE. model ExperimentalFactor value


ExperimentalStudy
   Keep typeCode(CD) and add designType (SC) and use DSET
   include systems like disease, pathway in definition
   change purposeStatement to description
   AI: to look at FuGE ExperimentalStudy


Experiment
   Keep designType
   AI: revisit Experiment attributes post planned and performed
   Use name, not title


ExperimentalItem
   Associate to Protocol
   Use just Quantity:PQ
   Remove name:ST (have typeCode)
Software
   Add licenseEffectiveDateRange (TS.DATE.FULL) accepted, check data type IVL TS
   need to create association between software and document.
   Add name:ST
   Add licenseType and remove openSourceFlag (BL):
   Not sure what the conclusion was on expirationDate (TS.DATE.FULL): The date
    (and time), assigned by the manufacturer, on which the product should not be
    used.
   Add Datatype = IVL<TS.DATE>
   Add
       o version(ST)
       o buildNumber (ST)
       o buildDate (TS.DATE )
       o licenseTypeCode (CD)
       o licenseEffectiveDateRange (IVL<TS.DATE>
       o licenseKey (ST)
       o Content(ED)


PointofContact
•   Use the defintion - Person or a department serving as the coordinator or focal point of
    an activity or program
•   Use attribute typeCode [vs. roleType]



DAY 2
Point of Contact Revisited
Discussion resumed with role class considerations.

POC is either an Organization or Person, but for multiple POCs each instance is either
an Organization or a Person.
Do we want Person POC and Organization POC to be constrained at this level? If yes,
need to be consistent throughout. Implication for Experimental Item.
POC has typecode (so it can be instantiated).

Discussion of structural and functional roles and the inclusion of an affiliation.
Figure 1. Point of Contact Proposal
POC associate with Organization and Person. Organization and Person each
associated with OrganizationalContact. To constrain PersonPOC sits between Person
and POC; Organization POC sits between Organization and POC;
OrganizationalContact is associated with POC.




Figure 2. Point of Contact Proposal
Same as Figure 1, but an Affiliate Class is added. It is associated to Person,
Organization and POC.
Lisa’s recommendation is to remove the red and green and use constraints. Make
note for lower layers. [Meaning simplest diagram is accepted: POC associate with
Organization and Person. Organization and Person each associated with
OrganizationalContact. Constraints to be added.]

Conclusion: POC as a class accepted and the Organizational Contact attributes should
be added to it. Include the constraints.
Parking Lot: POC affiliation (where is this person coming from).
AI: Experimental Item definitions revisit since they are not considering role class.
Cardinality will need to be checked. (Role Material, Software, Animal)


DocumentAuthor
Conclusion: Keep DocumentAuthor to Person association, but note there is no
connection of a person to organization. See parking lot.
Parking Lot: Affiliate (structural) may apply to DocumentAuthor


ExperimentalItem revisited
Discussion of roles being represented.
Quantity is an attribute or property for the association.
Consideration of overloading Material. Implementation may add a class, should this
be at the DAM level? How much you are using in an Experiment?
Suggest keeping Material and ExperimentalItem, add class Quantity (think of a
better name) with PQ.
How much of a Material is captured through Protocol – but is this plan? There is no
direct association between Material and Protocol for a materials list?
Experimental Item purpose is for quantity and to represent the role.
Perspective of aliquot or perspective of whole sample discussed. Material can be both
since it is recursive.
Material allows for composition, but there is no property for quantity.
Is quantity an issue for a DAM? Protocol is about reproducibility. You wouldn’t specify
rpm for a microfuge in a DAM.
typeCode captures the role the item plays.

Conclusion: ExperimentalItem - quantity is not at the DAM level.
AI: ExperimentalItem -update definition of typeCode:CD to: A code representing the
role an entity plays in an experiment.


Revisited Protocol
It is not an ExperimentalItem. Associations should be the same as
ExperimentalItem? No.
Conclusion: Have link from ExperimentalItem to Protocol and Experiment. Material
list for protocols can be captured through an experiment template – see experiment
typeCode.
Factor associated to Experiment vs. through ExperimentalItem.

AI: caught typos in definitions when reviewing Joyce’s comments, also have a few
items to discuss with her.

There is an association between Factor and Experiment and between Factor and
Study.
Conclusion: Leave as is.




Discussion with Joyce Hernandez.
ExperimentalStudy.
Anything deals with omics is well represented. You sSend omics data with a study.
Clinical trials reporting to FDA – is trial design represented using BRIDG, reusing LS
objects?
Joyce wants these aligned. If you have a study designed to look for new biomarkers
– you will have to explain to the FDA.
AI: take to BRIDG SCC to make their study as a specialization of our
ExperimentalStudy
typeCodes not bound
AI: add examples to show the LS ExperimentalStudy is the parent class

Experiment.
Wants the model to have overarching classes to associate to any omics.
The intention of the model is to be so generic that any type of experiment or assay is
captured – beyond omics. Linked to BRIDG through performed activity as is
gathering clinical data via performed activity.
AI: make the intention of the model explicit.


Subdomains
Discussion of utility of subdomains. Relationships has been the harder part.
Even if you decouple as subdomains – they are still maintained together.
Nano as an example – where you fit things in still the issue.
Q. What terms are the issue? For who?
Thinking is we can end up back to subpackages and logical groupings
Generate out views for use by whom?

Conclusion: Not enough views yet. Recommend tabling this activity
Will justify by capturing issues with entity naming especially as we go into new
activities.
ST or ED for BibliographicDesignation
Use String for BibliographicDesignation discussed. Some thinking that the use cases
for ED are not at the DAM level. Discussion of EndNote. Use formatCode.

Conclusion: Use ED for BiobliographicDesignation
AI: take a recommendation to BRIDG SCC to change the datatype of
DocumentVerisonBibliographicDesignation from ST to ED. By doing this, if you
specify the MIME type, you can specify the format. In ED you can also choose to use
string (it will be value).


Place
AD is a datatype and room is the smallest granularity you can get to.

Parking Lot: future development to include a container class

Discussion:
Container has a place and has a location
Quote of the Day:
“What do they call this in BRIDG?”
“I don’t care what they call it in BRIDG – I don’t speak BRIDG!”

Place is recursive and has: Id, p add, typeCode(CD), locator seq
There was a lot of wide ranging discussion.

Conclusion:
Existing Class and attributes
       Place (recursive)
       --------
       Identifier: DSET <II>
       typeCode:CD
       physicalAddress:AD
New attributes
       locatorValue: ST
       locatorType:CD

AI: Add locatorValue:ST and LocatorType:CD to Place
AI: XP to handle locator value and type in Place
Figure 3. Place Class
Place class is recursive. Existing Class (Place) with attributes: Identifier: DSET
<II>,typeCode:CD, physicalAddress:AD. New attributes are: locatorValue: ST and
locatorType:CD

Place Associations?
Existing association of Place To Document
AI: Add association of Place to Specimen
AI: Add association of Place to Container
AI: Add Container class as a Product and associated to Place

Container:
StatusCode
Status
AI: Lisa and Jenny to work with Mukesh.

Next Steps
Lisa to review notes and action items, make changes to the model and bring
questions to the group.

Next Meeting
October 21, 2010
11:00 am ET

Attendees

Konrad Rokicki, SAIC
Juli Klemm, NCI
Jenny Kelley, NCI
Grace Stafford, The Jackson Laboratory
Lisa Schick, ScenPro
Sue Pan, SAIC
Todd Stokes, GeorgiaTech
Robert Freimuth, Mayo Clinic
Jason Hipp, University of Michigan
Mukesh Sharma, Washington University
Baris Suzek, Georgetown University
Elaine Freund, 3rd Millennium

								
To top