TheJacksonLaboratory / ExperimentalModelSchema

Experimental Model Schema
https://thejacksonlaboratory.github.io/ExperimentalModelSchema/
MIT License
1 stars 0 forks source link

CohortMeasurements insufficiently defined #13

Open hansenp opened 1 year ago

hansenp commented 1 year ago

So far there is only a rudimentary definition of a message for measurements of cohorts that looks as follows:

message CohortMeasurement {
  string id = 1;
  int32 number_of_animals = 2;
  OntologyClass strain = 3;
  OntologyClass assay = 4;
  Sex sex = 5;
  double mean = 6;
  double minimum_value = 7;
  double maximum_value = 8;
  double standard_deviation = 9;
  double standard_error_of_mean = 10;
  double zscore = 11;
}

In my opinion, informations such as number_of_animals, strain or sex should not be part of this message. Rather, things like this should be part of the ExperimentalCohort element, which should be able to contain multiple CohortMeasurement instances. However, the ExperimentalCohort element has so far only been declared, but not defined. It would probably also be more flexible if lists were used for strain and sex, so that the element ExperimentalCohort can also be applied to mixed cohorts.

It is also not clear from the message what, when, how and where was measured. There is only the field assay, which is also undefined so far. In MPD, such information can be found in more or less unstructured form using the ID measnum.

The field standard_error_of_mean is redundant because it can be calculated from the fields standard_deviation and number_of_animals. Furthermore, in this context, a Z-score makes no sense to me. As I understand it, a Z-score indicates by how many standard deviations an observed value deviates from the mean. To which observed value does this Z-score refer here?