For the ilearn feature extractor, we take PseKNC, Z48bit, k-mer, and mismatch. PseKNC incorporates the k-tuple nucleotide composition. The Z48bit means Z_curve _48bit, the Z curve parameters for phase-independent tri-nucleotide frequencies. The mismatch profile also calculates the occurrences of kmers.
For the viral embedding, we selected the doc2vec feature extractor, and we applied doc2vec from gensim.
We selected bert, Albert, Roberta, GPT-2, Bart, and bioGPT for viral corpus extractors.
For the drug corpus, we selected bert, Albert, Roberta, GPT-2, Bart, and bioGPT as extractors.
For drug molecular figures, we selected EfficientNet_v21, EfficientNetb7, InceptionResNet, ResNest50, and ResNet101, and DinoVit as extractors.
For the drug network, we select role2vec, node2vec, GCN, and GAT as extractors.