Add more predictions to EpitopePrediction

[iskandr]

Possible groups of predictions (each mapping predictor name -> number):

• mhc_affinity_ic50
• mhc_affinity_score
• mhc_affinity_percentile_rank
• mhc_stability_hours
• mhc_stability_score
• mhc_stability_percentile_rank
• antigen_processing_score
• antigen_processing_percentile_rank
• pmhc_presentation_score
• pmhc_presentation_percentile_rank
• immunogenicity_score
• immunogenicity_percentile_rank

Also change wt_peptide_sequence to:

• closest_reference_peptide_sequence_in_any_protein
• closest_reference_peptide_sequence_in_same_protein

And calculate:

• edit_distance_to_closest_reference_peptide_in_any_protein
• edit_distance_to_closest_reference_peptide_in_same_protein
• edit_distance_to_closest_reference_peptide_in_any_protein_PMBEC
• edit_distance_to_closest_reference_peptide_in_same_protein_PMBEC

How to avoid duplicating every prediction for MT vs. WT-any vs. WT-same?

Possible API:

MHC affinity score:

epitope.predictions["mhc"]["affinity"]["score"]["NetMHCpan-4.1"]
epitope.predictions["mhc"]["affinity"]["score"]["MHCflurry-2.0"]
epitope.predictions["mhc"]["affinity"]["score"]["median"]
epitope.predictions["mhc"]["affinity"]["score"]["mean"]

Integrated Antigen Processing has one fewer level?

epitope.predictions["processing"]["score"]["MHCflurry-2.0"]