PACE

Welcome to PACE, a framework for comparing machine learning algorithms for the peptide-MHC binding problem. While many potential solutions to this problem have been published, a shared dataset and evaluation protocol have been missing, making it impossible to compare the various approaches. PACE (peptide algorithm comparison environment) fills that void by supplying a 16 HLA allele dataset and a structured evaluation protocol such that algorithms can be fairly compared.

As a secondary goal, PACE aims to provide useful utilities to facilitate the development of prediction algorithms, such as characteristics for alleles, utilities for implementing algorithms on common machine learning frameworks, etc.