TABLE 3.

Notations.

Notation	Meaning
Peptides and alleles
p	A peptide
a	An amino acid of a peptide sequence
$\mathcal{P}$	A set of peptides
$Q$	An allele
Binding
x/b	Original/normalized binding affinity for a peptide-MHC pair
l	Binding level for a peptide-MHC pair
Embeddings
e	Encoding vector of amino acid type
r	Embedding vector of each amino acid
o	Position embedding of each k-mer amino acids
R	Feature matrix for a peptide sequence
F G	Feature matrix for a padded peptide sequence (i.e., input of global kernel in $SpConvM$)
Parameters
S(·)	A scoring function
d e	Dimension of amino acid embedding
d f	Number of filters in convolution layer
d o	Dimension of position embedding $\mathit{o}$
d g	Number of global kernels in $\mathtt{SpConvM}$
d r	Dimension of hidden units $\mathit{c}$ in fully connected layer
k	Kernel size in convolutional neural layer
w	Attention weight learned in attention layer