This class is currently untested, but is a critical piece of infrastructure.
We should test the following (presumably with a mock for the actual API)
- it calls
_compute_msa when not all the required files are present.
_compute_msa calls the correct API_compute_msa correctly writes out the .a3m and .csv files and they contain the same output (see _validate_msa_cache_contents in msa.py for some good checks, on branch fix-path-serialization-bug if it hasn't yet been merged.- It correctly creates a cache directory, or uses one that is specified (test both)
- the hash function returns the correct values for the same ordered set of sequences
This class is currently untested, but is a critical piece of infrastructure.
We should test the following (presumably with a mock for the actual API)
_compute_msawhen not all the required files are present._compute_msacalls the correct API_compute_msacorrectly writes out the .a3m and .csv files and they contain the same output (see_validate_msa_cache_contentsin msa.py for some good checks, on branchfix-path-serialization-bugif it hasn't yet been merged.