**Summary:** - currently, PepSeqPred can only predict if a residue is or is not an epitope (i.e., 1 for yes, 0 for no) - to encode some level of uncertainty in the predictions, it would be helpful to add a 3rd uncertain class which is toggleable - for example, protein ABC123 could have the predicted epitopes: 000011122221111222001010101222000, where 0=not epitope, 1=yes epitope, and 2=uncertain **Done when:** 1. logic is cleanly integrated into existing training and inference scripts 2. logic is toggleable via the CLI scripts 3. appropriate unit, integration, and E2E tests are implemented and pass locally
Summary:
Done when: