Skip to main content
Fig. 3 | BMC Bioinformatics

Fig. 3

From: Amino acid encoding for deep learning applications

Fig. 3

Cluster heat-map of the pairwise distance among amino acids in different encoding schemes.a represents the learnt embedding space of a PPI-based model with eight dimensions. b represents the learnt embedding space of a peptide-HLA-II LSTM-based model with eight dimensions. c represents VHSE8 Matrix. d represents BLOSUM62. For the PPI-based model, the model was trained for 50 epochs using the full training dataset as described in the Materials and Methods section. For the RNN-based model, the model was trained for 3000 epochs using HLA-DRB1*15:01 data as described in Materials and Methods. For both cases, the weights of the embedding layer after training was used for visualizing the learnt embedding space. The special character zero, used for padding shorter sequences, was also included in the analysis

Back to article page