registration of statistics for characteristics of specimens at the specimen set level

[lzehl]

It would be great if we find a way of how to enable the specimen set schemas (subject group, tissue sample collection) to structurally identify the statistics of the characteristics of their individuals.

For example, currently a subject group can only structurally define the total number of individuals and that it contains male and female subjects, but not how many of those subjects are male or female.

At the moment we are solving this as a work around by adding this type of information unstructured in a string under "additional remarks" (for our example: "This group of 10 subject includes 4 males and 6 females.").

@openMetadataInitiative/openminds-developers Let's brain storm ideas how this can be solved structurally.

I see already two approaches (both with issues):

(1) allow for subject groups to be nested. So SG1 with 10 male + female subjects is the parent group of SG1.1 with 4 male subjects and SG1.2 with 6 female subjects. This might get nasty though...

(2) introduce a general "statistic/count" schema. SG1 has 10 subjects and links under biological sex to two "statistic/count" schemas that can state the attribute (male or female) and the count respective. Question here would be how to restrict to the correct attributes for the respective properties in the specimen set.

(3) keep sets as they are add an additional count property with the schema from (2)

(4) combine (1) with (3)

[lzehl]

@tgbugs it would be great to get you input here as well :)

[apdavison]

If we have few statistics per subject group, then nesting (option 1) is probably simplest. If we might have many statistics, especially if this would involve overlapping sub-groups (e.g. sex vs disease), then a SubjectStatistic schema would be simpler. Something like:

SubjectGroupStatistic
- count (integer)
- attribute (StudyTarget)

[lzehl]

Let's target this issue at last for v5.0 and maybe even postpone it to v6.0

[lzehl]

APPROACH A: SubjectGroup:

• identifier
• numberOfSubjects
• statistics [
  • count
  • attribute (controlled terms) ]

APPROACH B (closer to what already exists): SubjectGroup:

• identifier
• numberOfSubjects
• species [
  • count
  • attribute (species) ]
• biolgicalSex [
  • count
  • attribute (biolgicalSex) ]

[lzehl]

if really part of v5.0 will be decided later