Workshop OBI Ann Arbor 2012 May 30 - June 1

DATE 11 members signed up for a doodle survey. A few from the University of Michigan have expressed interests in attending the meeting. Dates: after discussion, the final date for the meeting is settled as May 30 (Wed), May 31 (Thur), and June 1 (Fri).

Workshop Venue The Raymond W.Waggoner Conference Room in The Molecular & Behavioral Neuroscience Institute (MBNI) University of Michigan 205 Zina Pitcher Place, Ann Arbor, MI 48109-5720

Teleconference Setup Instruction

(1) To make the voice clear and sound, please dial the following phone: The phone number that people will dial in for most of the three day setup is: 734-647-8166 Password: 5335

Note: One good way to make the dial-in call is through Skype phone. The use of Skype is only for calling in. We won't use Skype for the teleconference since Skype cannot support PowerPoint presentation and the internet linking quality is often poor. Instead, we will use Adobe connect. See the instruction below.

(2) Connect to the web conference system using Adobe connect:

First, make sure you have the latest version of Flash installed on your browser 1. Connect to: http://connectnow.acrobat.com/kjwill 2. Enter your name and click Connect (connecting as a guest) 3. Once the meeting room is open, you may click Start my Webcam in the upper right to share video. 4. If you are presenting, click Share my Screen at the proper time and choose to share "Desktop"   Others will see your screen. 5. If you want to leave the meeting, click the Meeting menu at the top and choose Leave Meeting (you may return at anytime by closing your browser and reopening it, then entering the above URL and logging in again.

Direction from Detroit Metropolitan Wayne County Airport ~30 minutes one way. Check Google Maps for direction

Transportation 1. Taxi costs is about $30. 2. To reserve a shuttle from customtransit(to use this service, you need to reserve first)

Lodging

Here is a list of recommended hotels:

-- Holiday Inn Near the University of Michigan (Recommended). Address: 3600 Plymouth Road, Ann Arbor, Michigan 48105 (About 10 minutes driving from the conference room) Check Google Maps for direction from the hotel to the conference room. Reservation Phone: Direct (734) 769-9800 or Toll-free: (800) 800-5560 URL: http://www.hiannarbor.com/ Rate: $81 per night. Note: This is a special rate designed for our conference. Please reserve it for this rate before May 16th. When you reserve, please mention the group of "Department of Microbiology" or "M4D" (for online reservation) (Contact: Oliver He). The hotel has shuttle to the Medical School every hour, and we can reserve special shuttle service for our group as well. This is the recommended hotel for the workshop.

-- The Campus Inn - Hotel Ann Arbor. Address: 615 E. Huron St., Ann Arbor, MI 48104 Phones: 800.666.8693; 734.769.2200 URL: http://www.campusinn.com/ Rates: $214-249 per night. Note: This hotel is about 10 minutes walking distance from the conference room.

-- Red Roof Inn Ann Arbor - Univ. of Michigan North. Address: 3621 Plymouth Rd, Ann Arbor, MI 48105 (About 10 minutes driving from the conference room) Phone: 734-996-5800 URL: http://www.redroof.com/reservations/property-detail.aspx?pid=045 Rates: $70-90 per night.

-- Marriott Courtyard Ann Arbor. Address: 3205 Boardwalk · Ann Arbor, Michigan 48108 USA Phone: 734-995-5900 URL: http://www.marriott.com/hotels/travel/arbch-courtyard-ann-arbor/

-- Ann Arbor Bed & Breakfast. Address: 921 E Huron Street Ann Arbor Michigan Tel: 734-994-9100 URL: http://www.annarborbedandbreakfast.com/

Food and Entertainment Breakfast, coffee and lunch are available at the MBNI's cafeteria (upstairs of the conference room). Maucus recommended Blue Nile: http://www.bluenilemi.com/bluenile/7/content.htm Glass House Cafe at Palmer Commons: http://www.glasshousecafe.net/ (close to meeting room) Angelo's Restaurant- Ann Arbor: http://www.angelosa2.com/home.htm (close to meeting room) More Restaurants in Ann Arbor: http://www.annarbor.com/restaurants/

Tentative Tasks for the workshop

Related slides/documents can be found under: https://obi.svn.sourceforge.net/svnroot/obi/trunk/docs/presentations/OBI%20workshop%20May%202012%20Ann%20Arbor/

Wednesday

9:00 - 12:00 - OBI core part I Use cases - (James O, 30-45 minutes) OBI core (JZ, BP)

1:00 - 5:00 - ECO

- QTT phenotypes design pattern (BP - 30 minutes) - OBI-ECO (JZ, MC, PRS - 2.5 hours) - hierarchy of specimen (JF, JZ - 1 hour)

Thursday 9:00 - 12:00 - data - Sort out data item/measurement datum (BP, PRS, get IAO folks to call in. BP to send out write-up, 2 hours) - Statistics representation in OBI (OH, ... - 1 hour)

1:00 - 5:00 - OBI core part II - continue core discussion (2 hours) - OBI release, which steps should be taken for OBI release process (JZ, CT; 2 hours)

Friday 9:00 - 12:00 - manuscript (JF, BP, JZ, CT, MH, LS) - grant (LS, CS)

Defer: - discussion of timing and modeling "domain knowledge" within studies (JF)

MEETING NOTES:

Wednesday:

OBI Core:

Trim taxonomy: Focus on organism / homo sapiens; core is not necessarily equivalent to OWL file

Remove upper level ontology terms - Make an OBI OBO – just the 13 terms

Check granularity (CRID/organism)

What is missing from core (see table 1)

Add links to other ontologies to CORE as special terms

How do things get into an investigation:

Acquisition ->Object acquisition – buy an antibody, rat

Enrollment is separate process and missing

Specimen creation - Draw blood from patient

Specimen creation -> Storing specimen (Biobank)

Specimen Creation -> Sample collection in the wild

Acquisition -> Data acquisition

Cellular component – add to core; excluding molecular entities

Protein complex – GO term; taking over by PROT; we will wait

ChEBI has proteins

Quality = child of specifically dependent continuant (PATO) needs to be incorporated into Core

Wait for:

IAO publication, symbol, file

ChEBI/PRO – molecular entities

Disease to be classified

AI for BP: Change name of “XYZ file” terms -> Information content entity that follows the XYZ format; do not add term “file”.

AI James, send out email: give people one week to review, and announce that we will go through the terms in the next calls (textual and logical definition). We will take James list on the wiki to go through, polish definitions, and at that point will argue for individual terms if they should be included or not.

Table 1 – changes / additions to Core:

Occurrent Process Add biological process Add child: disease course Planned process – add enrollment Continuant Dependent Continuant Generically dependent continuant Info Content Entity Document – remove publication Symbol – add question mark/more discussion Textual entity – add hypothesis textual entity Directive info entity – add enrollment specification? Specifically dependent continuant Realizable entity Disposition Function – add biological function (GO) ? Add measure function Role – add subject role Plan Independent Continuant Material entity Organism – slim down Add cell (Cell Type Ontology) Add gross anatomical part (CARO) Cellular component (from GO) – excluding molecular entities Site – change to geographical location (GAZ)

ECO / Evidence

Evaluator makes assertion (conclusion test entity) is about entity Assertion method = a means by which a statement is made about an entity Evidence is_an information content entity (not only data) used in interpreting data to draw the conclusion and support the conclusion

Hierarchy of evidence: Evidence/Similarity evidence/sequence similarity evidence/sequence alignment/blast/protein blast Plan to add defined classes to OBI so that one ECO term can map to one OBI term.

Action Philippe, Jie, Marcus: Plan is to start by adding assays to OBI which are needed to cover ECO, and focus on those assays which are (implicitly) well defined in ECO. Philippe has worked on assay definitions, and run those by Marcus before adding.

Phenotype questions/to incorporate

Is it phenotypic conclusion or conclusion or organism phenotype assessment Phenotypes can be comparative (this gene KO leads to longer tail than wild type) And (static) phenotypes can be about a “wild type” – bats are flying mammals

BP comment: Phenotype (as in the biological description) is outside of OBI scope, and the interested parties will work on it.

Christian: Evidence ontology = EVE

Similar to ECO/OBI work, with differences: assertion (information entity) connected to statement by asserts uses roles for dependent continuants: conclusion role/ premise role has bearer assertion act of inference ~ drawing a conclusion appears that assertions and certainly statements do not fit into BFO we will collaborate to work on one ontology for evidence.

AI BP: send Christian OWL file corresponding to the slides shown on OBI evidence

Sharing the excel sheet/google doc PRS created for qtt import We are going to stay at high level – avoid potential overly-detailed (and incorrect) assignment Use unnamed classes, eg CpG evidence: assay and utilized function of device (GpC island microarray)

AI for Marcus – use ECO terms with definitions and create equivalent assays in OBI Marcus will summarize process developed for QTT file

Discussion of Specimen:

Need something small that stands for something large Need large entity; study subject role is tied to conclusion

Definition: study subject role – inheres in thing that the conclusion is about Can be device, can be biological entity / organism / population Edge case: usability study – how well does the control panel work on this new device Possible to have different subjects within one investigation (RNA, rat, liver) Idea of sample coming from larger entity / population

Definition: material transformation process with objective to obtain a material entity for potential use in an investigation Either material sample / specimen or new term representing both

AI  call next week: Implement discussion results. Retire material sample / sampling process. Send complete proposal out to list.

Thursday:

Oliver: Statistics

We need input / output for each method, with assumptions, when the method is valid, etc. Oliver plans to add assumptions / rules for use of the method We need also the corresponding metadata to the data (e.g. is normal)

James Malone –> GenePattern pipeline Currently need expert knowledge to use GenePattern Having the added information will make the pipeline usable by non-experts

Comments:

part of relation does not apply to ICE Add covariate role for investigation AIs: BP adding straightforward terms to OBI; Oliver will add other missing terms to tracker

Quality of measurements –

Add responsiveness (response that you expect to change in response to treatment of known efficacy) Also can have quality of methods – assays related to responsiveness We can have measured / calculated values in OBI Reliability is a quality – BFO defines as inhering in independent continuants

Uses:

Reliability / validity / responsiveness – mostly applied to psychometric and physiological measures / qualities. Physiological – neuroimaging; volt signal from brain in response to finger tapping test (primates and people).

AI: - Responsiveness – Oliver will work with Marcy for use case

Add accuracy and precision

AI: Create Grouping terms –measure of dispersion; measures of test performance

Is the distinction direct vs indirect measurements? Questionnaire - Cannot directly measure mental state. Direct measurements do not need these terms, use measure of variance. Not direct vs indirect – indirect simply use second order conclusions.

AI Oliver: will submit a list of statistics terms to OBI track system. These terms can be listed together. (provide both textual and logical definitions of request terms) Jen and Philippe have offered to help this effort.

email from Bjoern: ICE for measured/predicted/planned / calibrated values
 * First, we will create a general class "value specification" to deal with data about lengths / mass / whatever, independent of them being measurements or not:

"value specification"=def: an information content entity that is about an entity of which a value can be measured and for which such measurements can be represented on a categorical or quantitative scale.

"scalar value specification"=def: a value specification that has two parts: a scalar and a unit

"temperature value specification"=def: a scalar value specification that is about a temperature quality and which has a "temperature unit" part

The tricky part in this is what the value specification is about. I favor allowing them to be about both specific instances and about classes. According to Alan, this can be expressed in OWL using "value", rather than "some": "value specification" is_about value entity


 * Second, we will create defined classes that indicate how a given value specification was created. For each of these classes, there is a 'planned process' of which they are a specified output, and which always includes an objective of how the value specification is intended to be used:

"measured datum" - a value specification that has been determined by a measuring process in which the value itself was examined by a process that is intended to give truthful results modulo a measurement error. Example: The value I read of my outside thermometer.

"predicted datum" - a value specification that has been determined by a prediction process in which an estimate of a value is made without examining the value itself. The value specification is intended to be close to the value a measurement process would produce modulo a prediction error. Example: The temperature predicted for tomorrow in the weather report

"planned value" - a value specification that is part of a plan specification, and which indicates what a value an entity should have if it were measured while executing a plan. Example: The temperature specified in "Incubate the cells at 37 Degrees Celsius for 24 hours"

"setting value" - a value specification to which a control on an instrument has been set. The instrument has been calibrated to achieve a certain value if it were measured when running the instrument. Example: The temperature control knob in my analog toaster oven. (which does *not* measure the actual temperature, but adjusts the current)


 * Third, if blessed by the BFO group, we will add roles to this. Every value specification that is not a measurement datum upon its creation gains a role (predicted value role / planned value role etc.) which is realized when a comparison between a prediction and a measurement is made. This would be elegant, completely consistent with the above and how we deal with roles in other places in OBI, but I would advocate not waiting for BFO to implement this, and go with the above for now which is clearly better than the mess we have in IAO right now.

AI BP: involve IAO people in discussion

Class with form unit and value About A or B or C or D A entity B plan to execute model followed by execute plan to test hypothesis C design of instrument / calibration (is realized by; in the future)

D Setting - will be similar Implied measurement process – test machine is working properly

Discussion of core

•	Educational tool; guide to understanding •	Covers use cases •	Solid definitions •	Hooks into other ontologies

Generalize data transformation to Information processing in core Update Documenting definition (BP) –  … by incorporating the specified input. Retire enrollment – use human subject enrollment; Human Subject enrollment - New definition: Process of obtaining informed consent from person or legal guardian of person who meets study design criteria to participant in the investigation Has_specified output has part some population and has part only human and not organism

cohort role - Add the constraint human population – bp has done this Import from OCRE: study population = collection or organisms formally enrolled in a study cohort population – study population followed over time arm population = cohort population with same intervention. Also eligible population

BP comment: Inspect terms from OCRE rather than import. Do this during the core term discussion.

Add Acquisition back into core

Acquisition/object acquisition/data acquisition – BP repaired definitions: process of gaining possession of a material entity/data/continuant

Add units to core? Alan suggested use QUDT in place of Unit Ontology (UO), have not made final decision yet.

Specimen CREATION objective/process -> Specimen COLLECTION objective/process Deprecate material sampling process / material sample – create term as part of conclusion; only then it will be clear if the objective of sampling a larger population has been achieved

Propose to deprecate ‘material sampling process’

OBI technique issues:

OBI release:

AI for Jie: send OBI desired features to Oort developers. The goal is to use Oort tool for OBI release. AI for Carlo: take care of next OBI release. The deadline of next release being out is mid July. Start end of June.

MIROET: Ultimately Oort will incorporate MIROET. We are going to fix the current issue with minimal efforts: 	Switch to use He group endpoint for SPARQL query 	Use the OntoFox input file provided by Carlo for NCBITaxon terms. AI: Jie

Deprecate terms: Keep OBO obsolete parent terms and send request to Protégé developers for adding folder feature. Add obsolete property as True to the obsolete terms. AI: Jie: update documentation

Friday:

OBI general paper:

Jen: presented her suggestions and proposal for the general paper updates. After good discussions, here is the plan: Jen will lay out a specific plan, which will first be discussed among people in today’s meeting. After it is agreed by the group of people, the plan will be submitted to the OBI email list for further discussion. AI: Jennifer

Funding opportunities:

Different opportunities were discussed. We would like to take any chance to get the principle OBI research efforts funded. One possibility is to develop a commercial business model and charge for service fee for using OBI in different applications, like LIMS system, data integration, database development, etc. This would potentially allow us to develop an SBIR Phase I/II proposal, however to do so we would need to have a clear product; it is uncertain if a “consulting” only SBIR is possible. We may also look for opportunities from different agencies including NIH and NSF, such as R41 funding. AI Chris: continue information gathering on SBIR and other opportunities

Phenotype discussion:

Chris brought up the need to model phenotype in OBI. OBI has ‘phenotypic assessment’. We changed this term to ‘comparative phenotypic assessment’ and modified the definition of the OBI term. In many cases, phenotypes reflect the differences from control or reference organism. However, there are cases that we don’t need comparison, for example, “red eye”.

It was difficult to achieve an agreement in the discussion. Different phenotype ontology efforts have been available. There is a Mammalian Phenotype Ontology developed by MGI. Marcus is now working on project to develop a microbial phenotype ontology. OBI now considers phenotype as a disposition or quality. Oliver indicated that phenotype might also mean processes. Alan suggested that we need to get broader participation and discussions.

Summary of General action items:

AI Bjoern: check if we can find new call time that works for Alan (done)

AI James O, send out email about core term list (done)

AI Bjoern, fix terms with 'file' in label

AI Marcus, Philippe, Jie: Add assays to OBI for ECO mapping

AI Bjoern, send OWL file modeling evidence to Christian B

AI Oliver, work with Marcy etc. on need for additional statistical classes

AI Bjoern, work with IAO folks on blessing implementation of measured vs. predicted datum

AI Jie: send OBI desired features to Oort developers. The goal is to use Oort tool for OBI release.

AI Carlo: take care of next OBI release. The deadline of next release being out is mid July. Start end of June.

AI Jie: Update MIREOT documentation

AI Jennifer: Send updated manuscript

AI Chris: Continue funding discussion

Attendees of the Ann Arbor OBI workshop - Bjoern Peters, La Jolla Institute for Allergy and Immunology, La Jolla, CA, USA - Chris Stoeckert, University of Pennsylvania School of Medicine, Philadelphia, PA, USA - Jie Zheng, University of Pennsylvania School of Medicine, Philadelphia, PA, USA - Jennifer M Fostel, National Toxicology Program, NIEHS, Durham, NC, USA - Marcus Chibucos, University of Maryland School of Medicine - Alan Ruttenberg, State University of New York - University at Buffalo, Buffalo, NY, USA - James Overton, The University of Western Ontario, Canada - Yongqun "Oliver" He, University of Michigan Medical School - Yu (Asiyah) Lin, University of Michigan Medical School - Zuoshuang "Allen" Xiang, University of Michigan Medical School - Rebecca Racz, University of Michigan - Marcelline (Marcy) Harris, University of Michigan - Alfred (Al) Hero (only May 30 and 31), University of Michigan - Frank J. Manion, University of Michigan - Philippe Rocca-Serra, University of Oxford (remotely)