Inherit annotations (Experimental factors) not available for annotation type Text

[Johan Vallon-Christersson]

The functionality that allow rawbioassays to inherit experimental factor annotations (see ticket:1133) does not seem to work when the annotation type has value type Text. No warning is issued that rawbioassays are missing value for experimental factor (perhaps no value is still considered as text) and no option to inherit annotations is then available (no click-able text to attempt to inherit annotations).

This problem could perhaps be allivated by implementing the suggestions in ticket:1374

[Nicklas Nordborg]

Hmmm.... the function to automatically inherit annotation builds on the function that enumerates the values that have actually been used. Since for a typical text annotation all values are different it doesn't make any sense to enumerate all values, and thus it is not known if values exists or not. The same is true for free-text 'string' type annotation (the difference is that a string can be max 255 characters and a text can be 64k characters long). By the way, I don't see that it makes any sense to use a free-text annotation type as an experimental factor. What is the use case?

[Johan Vallon-Christersson]

Replying to nicklas:

Hmmm.... the function to automatically inherit annotation builds on the function that enumerates the values that have actually been used. Since for a typical text annotation all values are different it doesn't make any sense to enumerate all values, and thus it is not known if values exists or not. The same is true for free-text 'string' type annotation (the difference is that a string can be max 255 characters and a text can be 64k characters long). By the way, I don't see that it makes any sense to use a free-text annotation type as an experimental factor. What is the use case?

The use case is a free-text annotation type used to store a 10 digit barcode. In the specific scenario groups of 8 labeled extracts are annotated with the same barcode. Naturally, one could have used an enumeration type annotation where all used barcodes are inputted as available annotation values. By using a free-text annotation type labeled extracts are simply annotated without first having to input the barcodes as annotation type values.

[Nicklas Nordborg]

It's better to use the 'string' annotation type if there is only 10 digits. Or maybe even better a 'long' numerical annotation type. The text annotation type can take up to 64k characters are really only needed if you expect texts that are longer than 255 characters. Database-wise it is also better to use string or numerical values since they can be handled faster.

But I have also found that my first comment is not completely true. The required information (if the annotation is missing on some raw bioassays) is gathered when generating the table with raw bioassays. It's just that the information is not used when populating the table with experimental factors. A slight redesign of the code should make it work.

[Nicklas Nordborg]

(In [5117]) Fixes #1394: Inherit annotations (Experimental factors) not available for annotation type Text