specimen: days_to_collection field

[pnrobinson]

This should be mapped to the time_of_collection field of the GA4GH Biosample message.

Check what data is coming for this. Probably use the same days to ISO 8601 function as for disease and individual.

[msierk]

Here's the definition of days_to_collection from GDC:

"The number of days from the index date to the date a sample was collected for a specific study or project."

The Biosample element time_of_collection is "Age of the proband at the time the sample was taken. RECOMMENDED."

So we need to use the index age of the subject + days from the index date = age of subject at time sample was taken.

I'm not sure how to get the age of the subject (line 58 of cda_biosample_factory.py)

[msierk]

@pnrobinson @ielis I have a question about the Phenopacket age element. The documentation says it should be age: iso8601duration: "P25Y3M2D", but in op_individual the parameter is described as ":param iso8601duration: age represented as an ISO 8601 Period". Is the element 'Age' or 'iso8601duration'?

https://phenopacket-schema.readthedocs.io/en/latest/age.html

[ielis]

Hi @msierk

OpIndividual is a convenience wrapper for constructing instances of Individual component of the Phenopacket Schema. The purpose of the wrapper is to simplify creation of the Protobuf object. The Python bindings generated by the Protobuf "compiler" are sometimes hard to work with, and OpIndividual simplify this.

However, with the benefit of hindsight, I think OpIndividual does not help too much. So, I think I'll remove it from the code base.

To create a well-formatted Age, either to use in time_at_last_encounter in Individual, in time_of_collection of Biosample, or elsewhere, we must first have a well formatted ISO8601 duration string (such as P25Y3M2D in your example), and then set it to time_element.age.iso8601duration field.

You may need to fight the Protobuf bindings to get it done though.. 😕

In the long term, I would like to create a Python library to simplify this. We have such a library in Java (Phenopacket tools, in particular the builder package), but Python needs one too.

Finally, I agree that we should compute the age at collection per your suggestion, and it is not yet clear to me how to make the subject's age available in the function. We'll have to work something out. Let's keep this issue open until we can resolve it.