Datasets (MHCflurry 2.0)

For the motifs and length distributions, we used the 0.01 (i.e. 1%) cutoff. The files also include results for other cutoffs.

• Motif raw data: mhcflurry.ba.frequency_matrices.csv.bz2
• Length distribution raw data: mhcflurry.ba.length_distributions.csv.bz2
• MHCflurry BA train data: mhcflurry.ba.train_data.csv.bz2.
• Extracted MHC positions (pseudosequences): mhcflurry.allele_sequences.csv