Skip to main content
. 2021 May 26;49(W1):W469–W475. doi: 10.1093/nar/gkab398

Figure 3.

Figure 3.

The procedure of Seq-RNN: The embedding layer first embeds an amino acid into a dense vector, BiLSTM then generates the hidden representation by capturing the context information of amino acids, and the max-pooling layer finally produces the overall representation.