Skip to main content
. 2022 Aug 10;9:928534. doi: 10.3389/fmolb.2022.928534

TABLE 1.

Summary of the key deep learning protein design methods discussed in this review, with their generation type and generative model type indicated by a *. ∼ in the structure design field suggests that some minor design coincides with sequence design. The design target of each method is also provided.

Method Generation type Generative model Design target
Sequence Structure VAE GAN Autoregressive
Grisoni et al. (2018) * * Antimicrobial peptides
Müller et al. (2018) * * Membranolytic anticancer peptides
PepCVAE * * Antimicrobial peptides
Hawkins-Hooker et al. (2021) * * Luciferase enzymes
ProteinGAN * * MDH-like enzymes
Greener et al. (2018) * * Metalloproteins
Gupta and Zou (2019) * * Antimicrobial peptides
Amimeur et al. (2020) * * Human antibodies
Ingraham et al. (2019) * * Non-specific
ProteoGAN * * Non-specific
ProteinSolver * * Non-specific
Anand et al. (2022) * * Non-specific
Ig-VAE * * Immunoglobulins
Anand et al. (2019) * * Non-specific
Tischer et al. (2020) * Inverted structure prediction model Non-specific
Anishchenko et al. (2021) * * Inverted structure prediction model Non-specific
Norn et al. (2021) * Inverted structure prediction model Non-specific