Hc/fix impossible schemes

[HenryCrosswell]

Before submitting a pull request (PR), please read the contributing guide.

Please fill out as much of this template as you can, but if you have any problems or questions, just leave a comment and we will help out :)

Description

What is this PR

• Bug fix
• Addition of a new feature
• Other

Why is this PR needed?
Since the PyRAT db is manually populated, there is a possibility that someone could have incorrectly entered genotypes for the offspring.

What does this PR do?
This adds a function to standardise.py that checks whether the offsprings genotype is a possible combination of the two parents, using the Mendelian ratios.

References

closes issue #10

How has this PR been tested?

Added new test to test_standardise, along with new test data which includes a raw file with impossible schemes, and a standardised file that asserts that those schemes have been removed.

Is this a breaking change?

No

Does this PR require an update to the documentation?

Not yet.

Checklist:

• The code has been tested locally
• Tests have been added to cover all new functionality
• The documentation has been updated to reflect any changes
• The code has been formatted with pre-commit

[K-Meech]

Thanks @HenryCrosswell - this is looking great! Your additions are filtering the impossible schemes really nicely, which will stop any incorrect data making it to the later processing stages.

I've put a few comments below - most are minor wording suggestions, but the main one is we need to make sure un-genotyped individuals can pass through the standardise process without erroring.

[K-Meech]

By adding with pytest.raises(TypeError), this test is allowed to pass even though it throws an error during standardise_pyrat_csv and never reaches pd.testing.assert_frame_equal. This means it is no longer testing that genotypes are standardised correctly.

You'll need to remove the with pytest.raises(TypeError) and rather add a fix in _remove_impossible_breeding_schemes instead. Otherwise, at the moment, any ungenotyped individuals that enter standardise_pyrat_csv will cause the processing to stop early (we need these ungenotyped individuals later in the historical stats part - see this issue: #8)

[HenryCrosswell]

I have addressed this issues, however it has removed an impossible scheme from the impossible genotype dataset. Causing issues in this tests - test_handling_ungenotyped_individuals_in_stats. I attempted to fix by removing and then fixing the standardised offspring. However it caused a tangle of issues. So I have left the test_data as is, leaving only the dataframe missmatch.
[left]: (8, 11)
[right]: (9, 11)

[K-Meech]

Thanks @HenryCrosswell - I've updated ID_011 and the corresponding tests on this PR: #50 and on GIN. Once that is merged, you should be able to merge main into this branch and GIN's master into your GIN branch and hopefully the tests will pass.

[K-Meech]

@HenryCrosswell #50 is merged now, so could you update your branches as mentioned above? If you open a PR on GIN I can take a quick look over that and merge it 👍

[K-Meech]

To keep rows with un-genotyped offspring, you could add something like this here:

# If the offspring are un-genotyped, we keep the row as there is no way
# of checking if the breeding scheme was valid
if pd.isna(genotype_offspring):
    return pop

Hopefully this should make the test_standardise_genotypes test pass, although I see that there is one remaining impossible breeding scheme in the test file I missed. You can change: pyrat-data-forbidden-genotypes.csv / standardised-data-forbidden-genotypes.csv ID-011 genotype_mother to wt_het on your GIN branch.

[HenryCrosswell]

Tackled all your comments, although i did it within vscode, so there might be some small differences - let me know if you want me to change anything else.