Release Neofox v0.5.0

Neofox v0.5.0 release notes

Major changes

• Update to NetMHCpan 4.1
• Update to NetMHCIIpan 4.0
• Integrate PRIME (Schmidt et al, 2021) as an additional feature
• Ambiguous MHC allele representations are resolved now using IMGT/HLA database
• Support for a more flexible mutation model not requiring the wild type sequence
• Generator rate for alternatively defined neoepitopes (Rech et al, 2018) is included for MHC-I and MHC-II
• Extended support for Python 3.6, 3.7 and 3.8

Minor changes

• The neoantigen unique identifier is removed from the tabular output
• The order of columns in the tabular output is changed to improve readability
• The affinity threshold is now a parameter (--affinity-threshold) with default value 500000 nM (ie: virtually disabled)
• Queries to the proteome are optimised

Bugfixes

• Best MHC alleles from MixMHCpred and MixMHC2pred are now normalized
• Logs in the file are now complete
• Avoid crashes with rare (ie: O, U) or non existing (ie: B, J, X, Z) amino acids

Known issues

• When the provided wild type sequence does not match the wild type in the proteome the best epitope may not overlap the mutation. This may occur when the wild type was derived from other proteome database than Ensembl GRCh37 or when the wild type sequence contains a germline mutation. This request takes care of the first point. Regarding the second point, the proteome database should be chosen more flexible. Internally, we will change to uniprot in the next release.
• Due to hits in the WT proteome search, a limited amount of neoepitopes or none may be left after this filtering step. For instance, this can lead to the fact that all features related to 9mers will be NA as no 9mer neoepitope candidate is present.
• Rare amino acids (ie: O and U) are returned as X by netMHCpand and netMHCIIpan