💡 Notes

• This is a list accompanying our preprint: https://www.biorxiv.org/content/10.1101/2022.08.31.505981v1 . We focus on deep learning methods for protein design released after 2018 (and mostly 2019). This table complements Table 1 in our manuscript.

• We curated this list manually and as such it might be incomplete. Please drop us an email or open an issue if you find we didn't describe your method correctly or it's missing.

• We order the methods by release date (preprint when available) and categorize them in four classes (for more details on these categories see our preprint, Figure 1 and text):

  • 1️⃣: 'Fixed-backbone' protein design; p(sequence|structure)
  • 2️⃣: Structure generation; p(structure)
  • 3️⃣: Sequence generation; p(sequence) or p(sequence|sequence*)
  • 4️⃣: Concomitant protein and sequence design. p(sequence and structure) (which can be constrained).

• Others before us have also done a fantastic work assembling deep learning methods for other protein-related problems, sometimes overlapping with this list. We link these lists here:

• • Sean Peldom Zhang's super comprehensive list on protein design methods using DL
  • Kevin Yang's list on ML methods for protein research
  • Christian Dallago & Sergey Ovchinnikov's lists on structure prediction methods and protein language models.
  • Simon Dürr and Gina El Nesr's list on inverse folding

• 💥 This work was recently highlighted in Nature

Contributors

• @noeliaferruz
• @sacdallago
• @michaelheinzinger

Class I: Protein Sequence design ("Fixed-backbone")

Methods in this class attempt to solve the classical protein design problem: Find an optimal sequence that adopts a pre-determined 3D structure.

Name	Architecture	Number of Parameters	User Input	Output	Training Dataset	Paper	Code	Release Month/Year
SPIN2	FNN	~105k	3D structure	sequence	1,532 X-ray structures	Paper	Code used to be here - no longer available	2018/02
SPROF	CNN-LSTM	-	3D structure	sequence	1,532 X-ray structures	Paper	Code Web Server	2019/08
Ingraham et al.	modified Transformer	>3k		sequence	CATH 4.2 40% sequences/structures	Paper	Code	2019/12
ProDCoNN	CNN	>28k	3D structure	sequence	Two datasets: ID90TR: 17,044; ID30TR: 9,135 sequences/PDB pairs	Paper	Reimplementation	2019/12
Anand et al.	CNN	-	3D structure	Amino acid and side chain conformation	53,414 CATH domain structures	Paper	Code	2020/01
DenseCPD	CNN	3M	3D structure	sequence	11,227 X-ray structures	Paper	Web server Reimplementation	2020/01
ProteinSolver	GNN	-	3D structure	sequence	72,464,122 sequences/adjacency matrices pairs	Paper	Code	2020/03
Norn et al.	CNN	N/A	distances, angles, and dihedrals for every pair of residues (trRosetta)	sequence	N/A	Paper	Code	2020/07
GVP-GNN	GVP	-	3D structure	sequence	CATH 4.2 40% sequences/structures	Paper	Code	2020/09
Fold2Seq	modified Transformer	-	3D structure	sequence	45,995 3D structures from CATH 4.2 filtered @ 100%	Paper	Code	2021/06
CNN_protein_landscape	CNN	>10M	3D structure	sequence	16,569 PDB chains	Paper	Code	2021/08
Orellana et al.	GCN	-	3D structures	sequence	CATH 4.2 40% sequences/structures	Paper	-	2021/11
ABACUS-R	Transformer	152M	3D structures	sequence	CATH 4.2	Paper	Code	2022/02
ESM-IF1	GVP-Transformer	142M	3D structure	sequence	16k X-ray structures + 1.2M AF2 predictions	Paper	Code	2022/04
TERMinator	GNN	-	3D structures	sequences	CATH 4.2 40% sequences/structures	Paper	-	2022/04
McPartlon et al.	modified Transformer	-	3D structures	sequences	37k X-ray structures from BC40	Paper	-	2022/04
MIF	Structured GNN	6.8M	3D structure	sequence		Paper	Code	2022/05
ProteinMPNN	MPNN	1.8M	3D structure	sequence	CATH 4.2 40% sequences/structures	Paper	Code Web Interface	2022/06
ProDESIGN-LE	Transformer + FNN	-	3D structure	sequence	5,867,488 residues from PDB40	Paper	-	2022/07
TIMED	CNN	3M	3D structure	sequence	32k structures from the PISCES server	Paper	Code	2022/08
PiFold	GNN	-	3D structure	sequence	-	Paper		2022/09

Class II: Structure generation

Methods in this class generate structures unconditionally or from a set of secondary structural conditions.

Name	Architecture	Number of Parameters	User Input	Output	Training Dataset	Paper	Code	Release Month/Year
64GAN	GAN	-	-	contact map (3D structure via ADMM)	427,659 contact maps	Paper	-	2018/12
Anand et al.	GAN	-	-	distance map (3D structure via CNN)	800,000 distance maps	Paper		2019/03
RamaNet	LSTM	>2k	-	A sequence of φ and ψ angles	607 helical structures	Paper	Code	19/06
DECO-VAE	VAE	-	Structures represented as graphs	contact graph (translatable to contact map)	>650,000 contact graphs	Paper	Upon request	2020/04
SCUBA	NC-NN	~20k	secondary structure motif	backbone	12,465 structures	Paper	Code	2022/02
Ig-VAE	VAE	-	-	protein backbone coordinates	10,768 individual immunoglobulin domains	Paper	Code	2022/02
GENESIS	VAE	-	secondary structure motif	contact map	40,726 backbones with remodeled loops	Paper	-	2022/03
ProtDiff & SMCDiff	EGNN	-	Optional: structural motif	coordinates	4,269 PDB structures	Paper	-	2022/06
Lai et al.	VAE	-	topology	protein backbone coordinates	CATH 4.2 40% sequences/structures	Paper	-	2022/07
ProteinSGM	SDE + RefineNet	-	optional: masked matrices	matrices describing distance and torsional angles	10,361 CATH 4.3 95% structures	Paper	-	2022/07
FoldingDiff	Transformer	-	-	internal angles	CATH 4.2 40% structures	Paper	Code	2022/09

Class III: Sequence generation

Methods in this class generate sequences usually from autoregressive language models, and can sometimes be conditioned.

Name	Architecture	Number of Parameters	User Input	Output	Training Dataset	Paper	Code	Release Month/Year
ProteinGAN	GAN	60M		sequence	16,706 MDH sequences	Paper	Code	2019/10
ProGen	Transformer	1.2B	Optional: sequence or function	sequence	280M sequences	Paper		2020/03
ProtXLnet	Transformer	409M	Optional: sequence	sequence	UniRef100	Paper	Code	2020/07
ProtXL	Transformer	562M	Optional: sequence	sequence	BFD100	Paper		2020/07
ProtElectra-Generator	Transformer	420M	Optional: sequence	sequence	Uniref100	Paper	Code	2020/07
ProtT5	Transformer	11B	Optional: sequence	sequence	BFD100	Paper	Code	2020/07
EVE	VAE		MSA	Sequence	3,219 MSAs extracted from UniRef100	Paper	Code	2020/12
DARK3	Transformer	110M	Optional: sequence	sequence	615,000 synthetic sequences	Paper	-	2022/01
ReLSO	Modified transformer	110M	sequence	sequence and predicted value for label	directed evolution datasets	Paper	Code	2022/02
ProtGPT2	Transformer	739M	Optional: sequence	sequence	UniRef50	Paper	Code	2022/03
RITA	Transformer	1.2B	Optional: sequence	sequence	UniRef100	Paper	Code	2022/05
Tranception	Transformer	700M	Optional: sequence	sequence	UniRef100	Paper	Code	2022/05
ProGEN2	Transformer	6.4B	Optional: sequence	sequence	Uniref90+BF30	Paper	Code	2022/06

Class IV: Sequence and structure design

Methods in this class generate sequences and structures concomitantly, and include hallucination methods and constrained generation (inpainting)

Name	Architecture	Number of Parameters	User Input	Output	Training Dataset	Paper	Code	Release Month/Year
Hallucination	CNN (trRosetta)	N/A	random sequence	sequence/structure	N/A	Paper	Code	2020/07
Constrained hallucination	CNN (trRosetta)	N/A	sequence/structure	sequence/structure	N/A	Paper	Code	2020/11
Constrained hallucination2	CNN (RoseTTAFold)	N/A	sequence/structure	sequence/structure	N/A	Paper	Code	2021/11
RFjoint	CNN (RoseTTAFold, finetuned)	N/A	sequence/structure	sequence/structure	Finetuned with 25% PDB version 02/2020 + 75 % AF2 structures	Paper	Code	2021/11
Protein Diffusion	Diffussion model	-	Secondary structure motif sketches	sequence/structure	53,414 3D structures (95% CATH 4.2 S95)	Paper	Code	2022/05
Roney	AlphaFold2	N/A	random sequence	sequence/structure	N/A	Paper	Code	2022/06