ProteinGAN: A generative adversarial network that generates functional protein sequences

ProteinGAN: A generative adversarial network that generates functional protein sequences
— Read on phys.org/news/2021-04-proteingan-adversarial-network-functional-protein.html

ProteinGAN: A generative adversarial network that generates functional protein sequences

by Ingrid Fadelli , Phys.org

Figure summarizing ProteinGAN’s training. Given a random input vector, the generator network produces a protein sequence, which is scored by the discriminator network by comparing it to natural protein sequences. The generator tries to fool the discriminator by generating sequences that will eventually look like real ones (the generator never actually sees real enzyme sequences). Credit: Repecka et al.