-
Notifications
You must be signed in to change notification settings - Fork 25
Meeting_notes_20100309
- Convened by: Bill
- Chaired by: Philippe
- Notes by:
- Where: Skype
- Attendees: Bill, Philippe, Alan, Melanie
QTT: Quick Term Templates
Background documents
- discussion on the IAO list
- submitted paper Overcoming the ontology enrichment bottleneck with Quick Term Templates Philippe Rocca-Serra, Alan Ruttenberg, Martin J. O’Connor, Patricia L Whetzel, Daniel Schober, Jay Greenbaum, Mélanie Courtot, Ryan R. Brinkman, Susanna Assunta Sansone, Richard Scheuermann, the OBI consortium and Bjoern Peters
- OBI website
- ICBO 2009 presentation
IAO call: topic QTT and clinical test:
Attendees: Bill Hogan, Melanie Courtot, Alan Ruttenberg, Philippe Rocca-Serra.
- modify analyte assay representation to accomodate reliance on unit as instance
relation should be modified from: has_unit some to: has_unit value
- lab test results may measure the same entity but will return results in possibly different units: to represent the range of possible units, use:
has_unit value {,..,}
3.mmol per liter [UO:300] is not available from OLS
*Action Item: Alan may have requested those and assumed they were added - he will check with UO
- Assay measuring an enzymatic activity rate/volume as proxy for Protein presence and quantity:
First issue: -> analyte assay are defined as ('is quality measurement of' some 'molecular concentration') this somewhat does not apply to enzymatic activity concentration even though it is used as a proxy for a molecular concentration.
Second issue: -> is it not a true analyte assay?
NOTE: post-call: this answer is probably 'yes',
-> Bill used UI/L which generically indicates a concentration.
these types of assay do not measure directly the amount of proteins rather, they measure the rate of conversion of substrate by protein assuming a proportional relation between this rate and the amount for functional protein present in the 'evaluant'
The unit needed is missing and would back measuring the amount of catalytic activity to a reference volume
*Action Item: submit to UO
Discussion about setting quality automatically upon selection of unit AR: not always possible
Option: have more general relation towards other types of entities
- Discussion about "normal ranges"
Issue: Should clinical assay representation include information about upper and lower measurement values considered as "normal"?
Defining 'normality values' is hard as those vary depending on species, physiological / pathological state (for reference, http://en.wikipedia.org/wiki/File:Blood_values_sorted_by_mass_and_molar_concentration.png) -> overly complex.
Key observation by Bill Hogan: " normal range is passed around with result of a test" so might be worth considering the following representation
Main modification is to changing the specific output of an assay model scalar measurement datum with range -> + extra template to generate subtype of normal value the issue is that a lab test typically carries in its result both the tested value and the upper and lower range of normal for that test instance
For evaluation purpose though, a QTT could be modified to test the feasibility of the following representation: "human normal blood sodium concentration value" is_a "normal value" has_measurement_value some {( >32, <38 )} has_measurement_unit mg/dL
Simplest case two extra columns to the template to allow reporting of (min , max) normal.
- Derived Measurements computed from Measurement values resulting from analyte assays:
Some tests as described in OBI-QTT-Analyte-Assay.xls document such "BUN/Cr ratio" or "albumin/globulin ratio" aren't really 'analyte assay', rather should be data transformation where the specific_inputs are measurement values from BUN and Creatinine analyte assays.
For simple ratios, a new qtt could be create where numerator and denominator are specified -> more discussion needed.