Another step on the road to accepting inputs such as 'EGFR T790M', I deleted a lot of the custom objects which assume that we always have access to a genomic variant (e.g. MutantEpitopePrediction) and made TopiaryPredictor methods for making predictions on protein sequences in the absence of variants.
Results are now returned in DataFrames, whose columns are documented in function docstrings. This isn't great for maintainability, but neither was having a sprawling mass of custom data objects.
I also deleted LazyLigandomeDict since we never use it in practice and propagating it through new code made refactoring even harder. In the future we should port over the reference proteome index from vaxrank.
Coverage decreased (-1.4%) to 86.224% when pulling 1125150065f5ec270119bca4283a35fbdf169264 on refactor-for-protein-sequences into 852c528e401a6e2708154ed544e2d629cf059f43 on master.
Another step on the road to accepting inputs such as 'EGFR T790M', I deleted a lot of the custom objects which assume that we always have access to a genomic variant (e.g.
MutantEpitopePrediction
) and madeTopiaryPredictor
methods for making predictions on protein sequences in the absence of variants.Results are now returned in DataFrames, whose columns are documented in function docstrings. This isn't great for maintainability, but neither was having a sprawling mass of custom data objects.
I also deleted
LazyLigandomeDict
since we never use it in practice and propagating it through new code made refactoring even harder. In the future we should port over the reference proteome index from vaxrank.