iskandr
commented
6 years ago Extension idea: if we return a long list of similar sequences for each epitope, then we can also run MHC binding predictors on them, since cross-reactive sequences will most likely also have to bind the same MHC.
Every epitope should be annotated with its nearest neighbor(s) in the reference proteome and the edit distance to those sequences. The edit function should only allow substitutions but not insertions or deletions, since TCRs are positionally sensitive (and the edit distance will most likely be used to approximate cross-reactivity).