fasta as input
abbyjerger opened this issue · comments
Currently the input test data must be in csv format. The pre-inference steps only allow for ResNet50 and ESM2 embeddings to be created from csvs. Similarly, although there is an inference_fasta.py script, it only creates the ESM2 embeddings from the fasta, not the ResNet50 embeddings.
It would be nice to use a csv or a fasta file to create the structure and sequence embeddings for the pre-inference step.
The fasta_to_csv function in utils.py is not currently adding the protein sequence to the 'Sequence' column in the output csv.