Sample name uniqueness check added to accucor data loader

[hepcat72]

Summary Change Description

Added:

• A raised exception for duplicate sample headers in the accucor data loader
• A raised exception when the study name column can't be found
• A raised exception when the header cells are empty
• A more descriptive exception when the number of column headers differ in the two sheets

It's notable that without the uniqueness check, the read_excel method silently modifies the column header and this results in a cryptic error stating "Could not find sample .1 in the database.", where the appended ".1" was the modification to the second occurrence of the duplicate header.

Files checked in: DataRepo/management/commands/load_accucor_msruns.py DataRepo/tests/test_models.py DataRepo/utils.py DataRepo/example_data/obob_maven_6eaas_inf_sample_dupe.xlsx

Affected Issue Numbers

• Resolves Check sample uniqueness in the accucor loader script #66
• Partially resolves MSRUN Data loading  #45

Code Review Notes

Note that adding mangle_dupe_cols=False followed by df.columns.duplicated() in the validation code would have been preferable, but it turns out that mangle_dupe_cols=False as an option to pandas.read_excel is not yet supported.

The uniqueness validation happens inside load_accucor_data.py since what is passed to the AccuCorDataLoader is the sheets and the check needs to be a different read of the sheets (with header=None). Ideally, this check would be done in AccuCorDataLoader.validate_dataframes, but since pandas.read_excel always mangles non-unique column headers, there's no way to do the check without a separate read of the file with headers turned off. I'm open to suggestions on reorganizing that check.

References:

• idea to read with header=None
• read_excel
• mangle_dupe_cols=False unsupported

Checklist

• All issue requirements satisfied (or no linked issues)
• Linting passes.
• Migrations created & committed (or no model changes)
• Tests implemented (or no code changes)
• All tests pass

[jcmatese]

I think this should work, and may catch some edge cases for user input.

[lparsons]

This PR catches duplicate sample names in the files, and provides a clear error message when found. For this reason, I think we should merge it.

I think it's worth noting, however, that if a file had duplicate sample names, I do not believe it would not have loaded before these changes since the sample names wouldn't be found in the database (they would have had an extra suffix added). The same is true for the extra check on columns without names and the test explicit check for STUDY_NAME column, (neither of which have explicit tests, btw).

[hepcat72]

Since this is all approved and I addressed all outstanding issues in the manner suggested, I will merge.