yutanagano / tidytcells Goto Github PK

• Implement this but don't make it the default behaviour
• Make it default in the future with version 3

Deprecate setup.py and move to pyproject.toml

An example of a valid HLA type that would fail to match the regex r'^(HLA-)?([A-Za-z/]+\d??)(*?([\d:]+(G)?(P)?)[LSCAQN]?)?$' is "HLA-DR1*13:01:01". This HLA type is valid because it follows the standard HLA naming conventions, but it would not match the regex because it contains two colons (:) between the digits and the "G" letter, while the regex only allows for one colon. The regex also only allows for one optional letter at the end of the string, but this HLA type has two letters, "G" and "P".

For a lot of use cases, you want to blanket apply tidytcells' standardization function to columns of a table, with some cells missing values. Currently, the standardize functions expect the first argument to be a string, so being passed Nones or nans triggers an exception. For now, users are expected to handle that themselves, but should there be an option to handle that quietly within tidytcells?

When applying tr.standardize, one may already know that the gene should be an alpha or beta chain (even if the input string/symbol does not make this clear). Sometimes, this may add crucial information that is the difference between a standardisation success/failure. It would be useful to add the option for the user to specify whether it is an alpha or beta chain they are standardising. (feedback courtesy of @mm523)

Currently when a non-functional V or J gene is detected, a warning message appears which is potentially confusing. The current warning states that the attempted fix is with the original allele. Instead one might simply warn the user that the gene is non-functional (but that no fix was done).