Alex Rubinsteyn edited sectionContent_Text_.tex  almost 8 years ago

Commit id: de37707ad381c2a5c0ab3daa5f9111b39169fd9a

deletions | additions      

       

The primary difference between NetMHC and NetMHCpan is that the former is an {\it allele-specific} method which trains a separate predictor for each allele's binding dataset, whereas the latter is a {\it pan-allele} method whose inputs are vector encodings of both the peptide and a subset of MHC molecule's primary sequence. The conventional wisdom is that NetMHC performs better on alleles with many assayed ligands whereas NetMHCpan is superior for less well-characterized alleles~\cite{Gfeller_2016}.  In this paper we explore the space between {\it allele-specific} and {\it pan-allele} prediction by imputing the unobserved values of peptide-MHC affinities for which we have no measurements and using these imputed values for pre-training of allele-specific binding predictors.  To fairly compare against NetMHC, NetMHCpan, and SMM on the same training and test samples we use the BD2009 and BLIND datasets from   \begin{itemize}  \item Perform cross-validation