The DOI matches its Crossref record; 77 of the 77 references it cites were found in a registry or an index; 1 of the 23 sources its claims cite were checked for support.
Large language models generate functional protein sequences across diverse families
Open accessAbstract
Deep-learning language models have shown promise in various biotechnological applications, including protein design and engineering. Here we describe ProGen, a language model that can generate protein sequences with a predictable function across large protein families, akin to generating grammatically and semantically correct natural language sentences on diverse topics. The model was trained on 280 million protein sequences from >19,000 families and is augmented with control tags specifying protein properties. ProGen can be further fine-tuned to curated sequences and tags to improve controllable generation performance of proteins from families with sufficient homologous samples. Artificial proteins fine-tuned to five distinct lysozyme families showed similar catalytic efficiencies as natural lysozymes, with sequence identity to natural proteins as low as 31.4%. ProGen is readily adapted to diverse protein families, as we demonstrate with chorismate mutase and malate dehydrogenase.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
32 Own result · 14 Own interpretation · 11 Not checked yet · 8 Own method · 4 No basis given · 1 Couldn’t confirm · accuracy not yet measured
The cited sources that could be checked couldn’t confirm or dispute this claim.
In the paper
“As in the crystal structure of L056, the predicted artificial structures roughly match known structures found in nature, Fig S11, including for low identity (<40%) artificial sequences.”
§ Large language models generate functional protein sequences across diverse families > Abstract
What the sources say
- Could not verifycites [14] Jumper, J. et al. Highly accurate protein structure prediction with AlphaFold. Nature 596, 583–589 (2021). sourceExperimental — accuracy not yet measured
The Abstract and Main section describe AlphaFold’s structure-prediction accuracy, including cases without a known similar structure. Our copy does not address whether predicted structures of functional artificial sequences resemble natural structures, including at below 40% sequence identity (searched: artificial sequences, L056, and 40% identity). (LLM, conf=0.77)
plaincorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Could not verify
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Our protein language model is trained by simply learning to predict the next amino acid for over 280 million protein sequences from thousands of protein families, without explicit biophysical modeling.”
§ Large language models generate functional protein sequences across diverse families > Abstract
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“After training, ProGen can be prompted to generate full-length protein sequences for any protein family from scratch, with a varying degree of similarity to natural proteins.”
§ Large language models generate functional protein sequences across diverse families > Abstract
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
None of the sources cited for this claim have been checked yet.
In the paper
“Our artificial lysozymes span the sequence landscape of natural lysozymes, Fig. 2a, across five families which contain diverse protein folds, active site architectures, and enzymatic mechanisms49,50.”
§ Large language models generate functional protein sequences across diverse families > Abstract
What the sources say
- Not checked yetcites [49] Baase, W. A., Liu, L., Tronrud, D. E. & Matthews, B. W. Lessons from the lysozyme of phage T4. Protein Sci. 19, 631–641 (2010). source
- Not checked yetcites [50] Kuroki, R., Weaver, L. H. & Matthews, B. W. A covalent enzyme–substrate intermediate with saccharide distortion in a mutant T4 lysozyme. Science 262, 2030–2033 (1993). source
plainn = 5- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Rests on an assumption: The five lysozyme families chosen because they lack disulfide bonds can be expressed in high-throughput cell-free systems.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“they demonstrate similar residue position entropies when forming separate multiple sequence alignments of natural and artificial proteins within each family, Fig. 2c.”
§ Large language models generate functional protein sequences across diverse families > Abstract
boostedn = 5- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“In the positive control set of 100 natural proteins, 72% were well-expressed based on chromatography peaks and band visualization.”
§ Large language models generate functional protein sequences across diverse families > Abstract
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Our artificial proteins match activity levels of natural proteins even at lower levels of sequence identity to any known natural protein, (Fig 3b, Fig. S5).”
§ Large language models generate functional protein sequences across diverse families > Abstract
plaincorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The representative natural proteins exhibited similar levels of functionality with 59% (53/90) of total proteins considered functional.”
§ Large language models generate functional protein sequences across diverse families > Abstract
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
In the paper
“None of the bmDCA artificial proteins exhibited a detectable level of functionality, Fig. S10”
§ Large language models generate functional protein sequences across diverse families > Abstract
What the sources say
- Not checked yetcites [7] Russ, W. P. et al. An evolution-based model for designing chorismate mutase enzymes. Science 369, 440–445 (2020). source
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Generated proteins exhibit similar conservation patterns to natural sequence libraries (Fig. 4ad).”
§ Large language models generate functional protein sequences across diverse families > Abstract
plaincorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- It may state two separate claims as one. 29
- Probably not a substantial claim. 16
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“ProGen is a 1.2 billion-parameter neural network trained using a publicly available dataset of 280 million protein sequences.”
§ Large language models generate functional protein sequences across diverse families > Abstract
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“we could readily measure the KM values for the purified apparent lower molecular weight species, where both L056 and L070 were highly active and had comparable Michaelis-Menten parameters to HEWL (Fig. 3e).”
§ Large language models generate functional protein sequences across diverse families > Abstract
hedged- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Among our artificial proteins, 73% (66/90) were functional and exhibited high levels of functionality across families, Fig. 3c.”
§ Large language models generate functional protein sequences across diverse families > Abstract
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Notably a small number of proteins, both within the natural and artificial proteins, were within an order of magnitude of HEWL, which was substantially more active than all negative controls.”
§ Large language models generate functional protein sequences across diverse families > Abstract
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Two proteins, L056 (max ID 69.6%) and L070 (max ID 89.2%) expressed well and incurred bactericidal activities towards the E. coli BL21(DE3) strain used during overnight induction at 16C.”
§ Large language models generate functional protein sequences across diverse families > Abstract
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The ProGen-generated proteins express equally well (72/100 total) across all bins of sequence identity to any known natural protein (maxID 40–90%), Fig 2e.”
§ Large language models generate functional protein sequences across diverse families > Abstract
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Despite having comparable enzymatic activities, L070 and L056 only share 17.9% sequence ID.”
§ Large language models generate functional protein sequences across diverse families > Abstract
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“We also found that there is no bias to location or structural element to the mutations that diverge L056 and L070 from their respective nearest sequence homolog in nature. Differing residues are instead uniformly distributed.”
§ Large language models generate functional protein sequences across diverse families > Abstract
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The KM values of both monomers were too weak to be measured using a heterogeneous, fluorescein-labeled Micrococcus lysodeikticus cell wall substrate (Molecular Probes EnzChek Lysozyme kit); however, both monomers were active using a pseudo-first order kinetic assay (Fig. S7).”
§ Large language models generate functional protein sequences across diverse families > Abstract
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Some mutations are even found within the active site cleft and within regions that influence conformational state (for L056).”
§ Large language models generate functional protein sequences across diverse families > Abstract
plainSingle case- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
None of the sources cited for this claim have been checked yet.
In the paper
“We determined a 2.5 Å resolution crystal of L056 (see Methods, Figure 3e). The global fold was similar to predictions, with a Cα RMSD of 2.9 Å from the backbone structure predicted by trRosetta and 2.3 Å RMSD from a WT T4 Lysozyme structure59,60.”
§ Large language models generate functional protein sequences across diverse families > Abstract
What the sources say
plainSingle case- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
In the paper
“bmDCA was unable to fit three out of the five lysozyme families, and exhibited 60% detectable expression (30/50 proteins) for the remaining two protein families.”
§ Large language models generate functional protein sequences across diverse families > Abstract
What the sources say
- Not checked yetcites [7] Russ, W. P. et al. An evolution-based model for designing chorismate mutase enzymes. Science 369, 440–445 (2020). source
plainn = 5- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Alignment with a structure featuring a covalently trapped substrate (PDB 148L) reveals that the active site cleft is well formed with the key catalytic residue Glu15 (Glu 11 in T4L) and key substrate binding residue Thr30 (Thr26 in T4L) correctly positioned.”
§ Large language models generate functional protein sequences across diverse families > Abstract
plainSingle case- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“These likelihoods showed a significant drop in AUC of 0.18 for CM (p<0.0001, two-tailed test, n=1617) and 0.08 for MDH (p<0.1, two-tailed test, n=56), as compared to fine-tuned ProGen when predicting if an artificial sequence should function.”
§ Large language models generate functional protein sequences across diverse families > Abstract
plaindirectn = 1,617 ·- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
In the paper
“On MDH function data, ProGen log-likelihoods had an AUC of 0.94 (Fig. 4f), which was better than ProteinGAN discriminator scores, with an AUC of 0.87 (p<0.1, two-tailed test, n=56).”
§ Large language models generate functional protein sequences across diverse families > Abstract
What the sources say
- Not checked yetcites [66] Vig, J. et al. BERTology meets biology: Interpreting attention in protein language models. In International Conference on Learning Representations (ICLR, 2020). source
plainn = 56 ·- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
In the paper
“On CM function data, ProGen log-likelihoods had an area under the curve (AUC) of 0.85, significantly better (p<0.0001, two-tailed test, n=1617) than bmDCA, which had an AUC of 0.78 (Fig. 4c).”
§ Large language models generate functional protein sequences across diverse families > Abstract
What the sources say
- Not checked yetcites [7] Russ, W. P. et al. An evolution-based model for designing chorismate mutase enzymes. Science 369, 440–445 (2020). source
plainn = 1,617- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“We purified 6 highly expressed proteins and found that they were all active, but with significantly lower Michaelis-Menten activities than HEWL or the previously generated artificial proteins L056 and L070 (Fig 3f).”
§ Large language models generate functional protein sequences across diverse families > Abstract
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Of the selected sequences, 78 out of the 89 (88%) expressed well and 24 out of the 78 (31%) were soluble (Supplemental Figure 8).”
§ Large language models generate functional protein sequences across diverse families > Abstract
plainn = 2- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“L013 and L038 expressed highly to inclusion bodies.”
§ Large language models generate functional protein sequences across diverse families > Methods > E. coli Expression of Lysozyme Variants
plaincorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“For lowID lysozyme A5 the above protocol was altered slightly to accommodate lower catalytic activity of these variants: reaction volumes were increased to 20 μL, plate was covered with an optically transparent seal (Microseal ® ‘B’ seal; BioRad) to mitigate sample evaporation, fluorescent reads were taken every 5 mins for 16 hours with 5 s of linear plate shaking prior to each measurement to min”
§ Large language models generate functional protein sequences across diverse families > Methods > Michaelis-Menten kinetics of lysozyme variants using fluorescein-labeled M. lysodeikticus cell wall
plaindirectSingle case- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“The MDH functional labels are binary, so no thresholding was needed to compute AUCs.”
§ Large language models generate functional protein sequences across diverse families > Methods > Evaluating ProGen on Other Protein Systems
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The resultant data were not described by a single exponential model but were described by a double exponential model (Eq. 6), likely due to the heterogeneity of the substrate, and all data were fit in IgorPro7.”
§ Large language models generate functional protein sequences across diverse families > Methods > Lysozyme kcat/KM extrapolation from pseudo first-order kinetic data
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“For induced cultures of L056 and L070, significant turbidity was observed in the spent media after cells were pelleted at 3,500 r.c.f. for 30 mins at 4°C.”
§ Large language models generate functional protein sequences across diverse families > Methods > E. coli Expression of Lysozyme Variants
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Two peaks were resolved for each variant and that harbored lysozyme activity against the fluorescein-labeled M. Lysodeikticus cell wall substrate (EnzChek kit; ThermoFisher).”
- Also stated in: § Purification of L056 and L070 from spent media · plain
“Two peaks were resolved for each variant and that harbored lysozyme activity against the fluorescein-labeled M. Lysodeikticus cell wall substrate (EnzChek kit; ThermoFisher) that corresponded to peaks observed in the first pool purification.”
§ Large language models generate functional protein sequences across diverse families > Methods > Purification of L056 and L070 from spent media
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
- Also stated in: § Purification of L056 and L070 from spent media · plain
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Significant translational non-crystallographic symmetry and differences with the search model resulted in maps that were initially hard to interpret.”
§ Large language models generate functional protein sequences across diverse families > Methods > Crystallization and structure determination of L056
plaindirectSingle case- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“These results indicate that ProGen generates protein sequences that not only can express well but also maintain enzymatic function for diverse sequence landscapes across protein families.”
§ Large language models generate functional protein sequences across diverse families > Abstract
hedgeddirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
- Rests on an assumption: Well-expression and solubility in the cell-free system indicate that the artificial proteins are structurally well-folded.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Taken together, L056 and L070 harbor potent catalytic activity and bactericidal capabilities that are comparable to HEWL, while diverging from their nearest known natural sequence by 53 and 18 amino acids respectively.”
§ Large language models generate functional protein sequences across diverse families > Abstract
plaincorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
- Rests on an assumption: HEWL is a suitable reference for native lysozyme catalytic activity.
What the paper did — it cites no source for this, so there is none to check. The quote shows it is in the paper.
In the paper
“The perplexity on a held-out test set was monitored and did not exceed training set perplexity throughout model training.”
§ Large language models generate functional protein sequences across diverse families > Methods > ProGen Training
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“While we do not expect our language model to generate proteins that belong to a completely different distribution or domain (for example creating a new fold that catalyzes an unnatural reaction), it can significantly expand the space of protein sequences from those sampled by evolution.”
§ Large language models generate functional protein sequences across diverse families > Abstract
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
- Rests on an assumption: The 280 million-sequence training set is diverse enough to learn a universal protein representation transferable across families.
What the paper did — it cites no source for this, so there is none to check. The quote shows it is in the paper.
In the paper
“The aggregated dataset was split into a training set of size 280 million and two test sets, an out-of-distribution test set (OOD-test) of size 100k from 20 protein families and a randomly sampled in-domain test set (ID-test) of size 1 million, that were held-out for training and used for evaluation.”
§ Large language models generate functional protein sequences across diverse families > Methods > Training Data Curation
plainn = 20- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. It may state two separate claims as one.
None of the sources cited for this claim have been checked yet.
In the paper
“In sum, ProGen’s model likelihoods are better aligned with experimentally-measured assay data on two diverse protein datasets—chorismate mutase and malate dehydrogenase—than the sequence generation methods from original studies specifically tailored for these families.”
§ Large language models generate functional protein sequences across diverse families > Abstract
What the sources say
- Not checked yetcites [7] Russ, W. P. et al. An evolution-based model for designing chorismate mutase enzymes. Science 369, 440–445 (2020). source
- Not checked yetcites [66] Vig, J. et al. BERTology meets biology: Interpreting attention in protein language models. In International Conference on Learning Representations (ICLR, 2020). source
plaincorrelationaln = 1,617- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“These results indicate that both components of our training strategy—initial training on the universal sequence dataset and fine-tuning on the protein family of interest—contribute significantly to final model performance.”
§ Large language models generate functional protein sequences across diverse families > Abstract
hedgeddirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
What the paper did — it cites no source for this, so there is none to check. The quote shows it is in the paper.
In the paper
“For robustness, we trained three discriminators using different random seeds.”
§ Large language models generate functional protein sequences across diverse families > Methods > Lysozyme Sequence Selection
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
What the paper did — it cites no source for this, so there is none to check. The quote shows it is in the paper.
In the paper
“Proteins that generated fluorescence one standard deviation above the maximum fluorescence of any negative control were considered functional.”
§ Large language models generate functional protein sequences across diverse families > Abstract
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The resultant data were not described by a single exponential model but were described by a double exponential model (Eq. 6), likely due to the heterogeneity of the substrate, and all data were fit in IgorPro7.”
§ Large language models generate functional protein sequences across diverse families > Methods > Lysozyme kcat/KM extrapolation from pseudo first-order kinetic data
hedgedconditional- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
- Rests on an assumption: The average molecular weight of the fluorescein-labeled M. lysodeikticus cell wall substrate is unknown and the substrate is heterogeneous.
What the paper did — it cites no source for this, so there is none to check. The quote shows it is in the paper.
In the paper
“We also ensured that no two generated samples share over 80% identity with each other.”
§ Large language models generate functional protein sequences across diverse families > Abstract
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim.
What the paper did — it cites no source for this, so there is none to check. The quote shows it is in the paper.
In the paper
“After measuring the AUC of each model for each dataset, we used bootstrapping to compute the statistical significance of the difference in AUC of fine-tuned ProGen vs. the reference method (bmDCA and ProGen ablations for CM, ProteinGAN and ProGen ablations for MDH).”
§ Large language models generate functional protein sequences across diverse families > Methods > Evaluating ProGen on Other Protein Systems
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim.
What the paper did — it cites no source for this, so there is none to check. The quote shows it is in the paper.
In the paper
“The fine-tuning dataset was clustered at 80% sequence identity and 10% of the clusters were held-out as a validation set for hyper-parameter optimization and stopping criteria.”
§ Large language models generate functional protein sequences across diverse families > Methods > Lysozyme Generation
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.
In the paper
“We also applied thresholding at 0.42 norm relative enrichment (r.e.) to obtain binary labels for CM function, which roughly corresponds to the cut-off point between two modes that exist in CM function data, to be used for ROC curves, following the original study7.”
§ Large language models generate functional protein sequences across diverse families > Methods > Evaluating ProGen on Other Protein Systems
What the sources say
- Not checked yetcites [7] Russ, W. P. et al. An evolution-based model for designing chorismate mutase enzymes. Science 369, 440–445 (2020). source
hedgedcorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
What the paper did — it cites no source for this, so there is none to check. The quote shows it is in the paper.
In the paper
“We performed structure prediction without an MSA as input so as to not heavily bias the structure prediction toward a known natural mode.”
§ Large language models generate functional protein sequences across diverse families > Abstract
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim.
The paper relies on this without support.
The five lysozyme families chosen because they lack disulfide bonds can be expressed in high-throughput cell-free systems.
“Within the lysozyme clan, we chose to work with five families that lack disulfide bonds as to be amenable to high-throughput cell-free expression: Phage lysozyme, Pesticin, Glucosaminidase, Glycoside hydrolase, and Transglycosylase.”
The paper relies on this without support.
Well-expression and solubility in the cell-free system indicate that the artificial proteins are structurally well-folded.
“These results indicate that ProGen can generate artificial proteins that are structurally well-folded for proper expression as compared to a batch of natural proteins”
The paper relies on this without support.
The 280 million-sequence training set is diverse enough to learn a universal protein representation transferable across families.
“Trained in this unsupervised manner from a large, varied protein sequence database, ProGen learns a universal, domain-independent representation of proteins that subsumes local and global structure motifs”
The paper relies on this without support.
HEWL is a suitable reference for native lysozyme catalytic activity.
“Hen egg white lysozyme (HEWL), a naturally-evolved exemplar protein, was measured as positive control”
The paper relies on this without support.
Model likelihoods for GANs are intractable.
“Since model likelihoods for GANs are intractable, we used discriminator scores corresponding to the probability at which the ProteinGAN discriminator predicted each sample was real as a ProteinGAN model score for each MDH sequence.”
The paper relies on this without support.
The average molecular weight of the fluorescein-labeled M. lysodeikticus cell wall substrate is unknown and the substrate is heterogeneous.
“(the average molecular weight of the fluorescein-labeled M. Lysodeikticus cell wall substrate was unknown and likely heterogeneous).”
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“synthetic proteins optimized for specific families of lysozymes using ProGen have exhibited catalytic efficiencies comparable to their naturally occurring counterparts with low sequence identity.”
From Large language models (LLMs): survey, technical frameworks, and future challenges (2024) · cites this paper as Madani et al. 2023
Experimental — accuracy not yet measuredThe cited source says what the claim attributes to it.
In the citing paper
“AI models have already been used to design new CRISPR enzymes,53 functional proteins,128 and even entire prokaryotic genomes.51”
From How to build the virtual cell with artificial intelligence: Priorities and opportunities (2024) · cites this paper as [128]
The cited source says what the claim attributes to it.
In the citing paper
“The success of LLMs has revolutionized AI learning in natural language and image processing [75] and has been extended into the learning of biological sequence datasets [80].”
From Challenges in AI-driven Biomedical Multimodal Data Fusion and Analysis (2025) · cites this paper as [80]
The cited source says what the claim attributes to it.
In the citing paper
“the ProGen (Madani et al. 2023) language model employs deep learning techniques to produce protein sequences with predefined functionalities.”
From Large language models (LLMs): survey, technical frameworks, and future challenges (2024) · cites this paper as Madani et al. 2023
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ Large language models generate functional protein sequences across diverse families > Abstract
“a state-of-the-art transformer-based conditional language model trained only with evolutionary sequence data”
§ Large language models generate functional protein sequences across diverse families > Methods > ProGen Training
“The model was initialized with pretrained weights of CTRL28, which was trained on an English language corpus.”
Citations
Cites
Adam: A Method for Stochastic Optimization
Kingma, D. P., & Ba, J. (2014) · arXiv · doi:10.48550/arxiv.1412.6980 · Cited by 98,946
LeCun, Y., Bengio, Y., & Hinton, G. (2015) · Nature · doi:10.1038/nature14539 · Cited by 84,126
Highly accurate protein structure prediction with AlphaFold
Jumper, J., Evans, R., Pritzel, A., Green, T., Figurnov, M., Ronneberger, O., Tunyasuvunakool, K., Bates, R., Žídek, A., Potapenko, A., et al. (2021) · Nature · doi:10.1038/s41586-021-03819-2 · Cited by 43,764
Gene Ontology: tool for the unification of biology
Ashburner, M., Ball, C. A., Blake, J. A., Botstein, D., Butler, H., Cherry, J. M., Davis, A. P., Dolinski, K., Dwight, S. S., Eppig, J. T., et al. (2000) · Nature Genetics · doi:10.1038/75556 · Cited by 43,206
Features and development of Coot
Emsley, P., Lohkamp, B., Scott, W. G., & Cowtan, K. (2010) · Acta Crystallographica Section D Biological Crystallography · doi:10.1107/s0907444910007493 · Cited by 29,915
Language Models are Few-Shot Learners
Brown, T. B., Mann, B., Ryder, N., Subbiah, M., Kaplan, J., Dhariwal, P., Neelakantan, A., Shyam, P., Sastry, G., Askell, A., et al. (2020) · arXiv · doi:10.48550/arxiv.2005.14165 · Cited by 23,372
Phaser crystallographic software
McCoy, A. J., Grosse-Kunstleve, R. W., Adams, P. D., Winn, M. D., Storoni, L. C., & Read, R. J. (2007) · Journal of Applied Crystallography · doi:10.1107/s0021889807021206 · Cited by 20,612
Kabsch, W. (2010) · Acta Crystallographica Section D Biological Crystallography · doi:10.1107/s0907444909047337 · Cited by 16,463
Deep Contextualized Word Representations
Peters, M., Neumann, M., Iyyer, M., Gardner, M., Clark, C., Lee, K., & Zettlemoyer, L. (2018) · Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long Papers) · doi:10.18653/v1/n18-1202 · Cited by 11,983
Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer
Raffel, C., Shazeer, N., Roberts, A., Lee, K., Narang, S., Matena, M., Zhou, Y., Li, W., & Liu, P. J. (2019) · arXiv · doi:10.48550/arxiv.1910.10683 · Cited by 10,929
Research works printed but not linked: 13
- Vaswani, A. et al. Attention is all you need. In 31st Conference on Neural Information Processing Systems (NIPS, 2017).
- Devlin, J., Chang, M.-W., Lee, K. & Toutanova, K. BERT: Pre-training of deep bidirectional transformers for language understanding. In Proceedings of North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT, 2019).
- Zellers, R. et al. Defending against neural fake news. In 33rd Conference on Neural Information Processing Systems (NeurIPS, 2019).
- Radford, A., Narasimhan, K., Salimans, T. & Sutskever, I. Improving language understanding by generative pre-training. Preprint at https://s3-us-west-2.amazonaws.com/openai-assets/research-covers/language-unsupervised/language_understanding_paper.pdf (2018).
- Dathathri, S. et al. Plug and play language models: a simple approach to controlled text generation. In 8th International Conference on Learning Representations (ICLR, 2020).
- Pettit, L. D. The IUPAC stability constants database. Chem. Int. 28, 14–15 (2006).
- Bengio, Y., Ducharme, R., Vincent, P. & Janvin, C. A neural probabilistic language model. J. Mach. Learn. Res. 3, 1137–1155 (2003).
- Goyal, K., Dyer, C. & Berg-Kirkpatrick, T. Exposing the implicit energy networks behind masked language models via metropolis–hastings. In 10th International Conference on Learning Representations (ICLR, 2022).
- Alley, E., Khimulya, G., Biswas, S., AlQuraishi, M. & Church, G. M. Unified rational protein engineering with sequence-based deep representation learning. Nat. Methods 16, 1315–1322 (2019). · DOI · Paper page
- Pascanu, R., Mikolov, T. & Bengio, Y. On the difficulty of training recurrent neural networks. In Proc. 30th International Conference on Machine Learning (eds. Dasgupta, S. & McAllester, D.) 1310–1318 (PMLR, 2013).
- Srivastava, N., Hinton, G., Krizhevsky, A., Sutskever, I. & Salakhutdinov, R. Dropout: a simple way to prevent neural networks from overfitting. J. Mach. Learn. Res. 15, 1929–1958 (2014).
- Holtzman, A., Buys, J., Du, L., Forbes, M. & Choi, Y. The curious case of neural text degeneration. In 8th International Conference on Learning Representations (ICLR, 2020).
- Goodfellow, I. J. et al. Generative adversarial networks. In 28th Conference on Neural Information Processing Systems (NIPS, 2014).
Cited by
Sorted by citations
Items 701–800 of 1,012
Biophysics-based protein language models for protein engineeringPreprint
Gelman, S., Johnson, B., Freschlin, C. R., D'Costa, S., Gitter, A., & Romero, P. (2024) · bioRxiv · doi:10.1101/2024.03.15.585128 · Cited by 2
“Predictive models can guide searches over the sequence-function landscape to enhance natural proteins or design new proteins [6, 31, 32].”
Evaluation of ChatGPT's Usefulness and Accuracy in Diagnostic Surgical Pathology.Preprint
Guastafierro, V., Corbitt, D. N., Bressan, A., Fernandes, B., Mintemur, O., Magnoli, F., Ronchi, S., Rosa, S. L., Uccella, S., & Renne, S. (2024) · medRxiv · doi:10.1101/2024.03.12.24304153 · Cited by 2
Deep-Learning Uncovers certain CCM Isoforms as Transcription Factors.
Croft, J., Gao, L.-Y., Sheng, V., & Zhang, J. (2024) · Frontiers in Bioscience · doi:10.31083/j.fbl2902075 · Cited by 2
Doron, G., Genway, S., Roberts, M., & Jasti, S. (2023) · arXiv.org · doi:10.48550/arxiv.2312.12482 · Cited by 2
“As discussed in the previous section, synthetic data applications need to be approached with caution, but the potential future value is significant.”
BackgroundEnhancing Power Grid Model Generation With Large Language Models via Progressive Fine-Tuning
Deng, K., Zhou, Y., Yang, Y., Yang, X., Dong, W., & Guo, Q. (2026) · IEEE Transactions on Power Systems · doi:10.1109/tpwrs.2026.3723876 · Cited by 1
“LLMs have demonstrated remarkable versatility across diverse domains, including functional protein sequence generation [12], chemical knowledge question answering [13], and mathematical problem solving [14].”
DESIGN: a framework for generative AI-enabled crop molecular design-driven breeding
Xu, F., Fu, R., Cheng, Q., Jiang, S., Yan, J., & Wang, X. (2026) · Science China Life Sciences · doi:10.1007/s11427-025-3406-3 · Cited by 1
Fu, L., Lv, Y., Chen, F., Zhao, S., Lin, C., & Karimi-Maleh, H. (2026) · TrAC Trends in Analytical Chemistry · doi:10.1016/j.trac.2026.119022 · Cited by 1
What deep learning can and cannot (yet) do for enzyme design
Ahr, J. A., de Waal, M. M., Cotet, T., Howell, D. C., & Bunzel, H. A. (2026) · Current Opinion in Chemical Biology · doi:10.1016/j.cbpa.2026.102770
Training Large Language Models for Small-Molecule Design with Synthetic Task ScalingPreprint
Hu, F., Chennakesavalu, S., Wang, Z., Suriana, P., Vani, B., Shmilovich, K., Chuang, K., & Grambow, C. A. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2609.04735 · Cited by 1
“…on small-molecule design (Narayanan et al., 2025; Chennakesavalu, Shmilovich, et al., 2026; Chennakesavalu, Hu, et al., 2025) , protein design (Madani et al., 2023; Widatalla et al., 2024; Fallahpour et al., 2026; Ibarraran et al., 2026) , and biology (Istrate et al., 2025; Anthropic, 2026a;…”
De novo design of ligand binding proteins using large language models alonePreprint
Kim, N. H., Hatstat, A. K., Jo, H., Wu, Y., & DeGrado, W. F. (2026) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.64898/2026.09.02.748987
Creating bottom-up RNA transfer vehicles from synthetic protein assemblies
Schuhmacher, M. K., Gruber, C., Lang, C. M. R., Ruijpers, R., Mazneykova, L., Beinsteiner, B., Krus, A., Tremmel, B., Reinhardt, F., Kadletz, K., et al. (2026) · Nature · doi:10.1038/s41586-026-10952-3 · Cited by 1
Programmable biodegradation: Lipase-driven microplastic degradation via AI and omics
Majumder, D., Dutta, A., Lahiri, D., & Nag, M. (2026) · Preparative Biochemistry & Biotechnology · doi:10.1080/10826068.2026.2720085 · Cited by 1
He, T., Zhang, S., Chen, J., Zheng, X., Zhao, J., Li, H., Mao, J., Luo, J., Zheng, J., Ma, D., et al. (2026) · iMetaOmics. · doi:10.1002/imo2.70133 · Cited by 1
Gu, J., Xu, Y., Sun, X., & Nie, Y. (2026) · ACS Synthetic Biology · doi:10.1021/acssynbio.6c00336 · Cited by 1
Enzyme Engineering: From Classical Strategies to AI-Driven Biocatalyst Design
Ullah, M., Rizwan, M., Andoh, V., Raza, A., Naeem, M., Waheed, M. I., & Chen, Y. (2026) · Journal of Agricultural and Food Chemistry · doi:10.1021/acs.jafc.6c04272 · Cited by 1
Qian, Z., Zhou, Y., Lin, R., Hong, Y., Xu, Y., Li, S., Gao, J., Zhu, Q., & Zheng, H. (2026) · Journal of Artificial Intelligence Research and Innovation · doi:10.29328/journal.jairi.1001022 · Cited by 1
“In combination with pretrained protein language models, generative frameworks greatly expand the accessible peptide landscape and support the design of bioactive peptides with tailored inhibitory pro(cid:976)iles [64].”
Basit, A. (2026) · Journal of Molecular Modeling · doi:10.1007/s00894-026-06846-w · Cited by 1
Chang, S., Lu, D., Jin, Y., He, Y., Zhao, Y., Xie, X. J., Guo, J., Xie, W., & Tao, Y. (2026) · Clinical Epigenetics · doi:10.1186/s13148-026-02194-x · Cited by 2
“…published in Nature Biotechnology in 2023 by Madani et al. and trained on 280 million protein sequences across more than 19,000 protein families [73], is a 1.2-billion-parameter autoregressive protein generation model; artificially generated lysozymes retain catalytic efficiency comparable to…”
InfluentialGatlin, W., Ludwick, M., Turano, L., Foley, B., Riedlová, K., Škrhák, V., Novotný, M., Hoksza, D., & Verkhivker, G. M. (2026) · Protein Science · doi:10.1002/pro.70714 · Cited by 1
Zhang, R., Zhang, Y.-Y., Deng, Z., Wang, P.-T., Fan, L., Guo, F., Luo, H.-Y., Wang, Y., Tu, T., Yao, B., et al. (2026) · Bioresource Technology · doi:10.1016/j.biortech.2026.135309 · Cited by 1
AI-Powered Synthetic Biology: Current Situation, Challenges, and Future Perspectives
Şahin, İ. O., Gökçe, N., Yildirim, D. T., Yildirim, A. B., Akalin, H., Martin, D. K., Beccari, T., Vicente, O., Radecka, I., Tchuenbou-Magaia, F. L., et al. (2026) · The EuroBiotech Journal · doi:10.2478/ebtj-2026-0015 · Cited by 1
“…language model trained on hundreds of millions of protein sequences can produce engineered enzymes with catalytic activity comparable to natural homologues, even when the engineered sequence shares only about thirty percent of its amino acids with any known protein (Madani et al., 2023).”
Plant synthetic biology: From modular design to artificial intelligence-driven engineering
Guo, Q., Kong, C., Zhang, J., Cao, P., Fu, S., & Wang, B. (2026) · Model Organisms Research · doi:10.1016/j.mores.2026.100008 · Cited by 1
Ogungbite, A. B., & Sibiya, M. (2026) · CPT Pharmacometrics & Systems Pharmacology · doi:10.1002/psp4.70272 · Cited by 2
“In practice, we anticipate hybrid stacks: KG/GNN backbones for structure and causality, with foundation models handling literature synthesis, hypothesis drafting, and protocol generation [10, 12, 34, 42, 52, 87].”
InfluentialBisardi, M., di Bari, L., Rossi, S., Buda, K., Zamponi, F., & Weigt, M. (2026) · doi:10.48550/arxiv.2606.29529 · Cited by 1
“In this context, recent protein language models have demonstrated the ability to capture broad evolutionary and functional relationships across diverse sequence spaces, enabling the generation of proteins with novel or altered functions [7, 166, 167].”
Two-Stage Fine-Tuning for Protein Sequence Generation with Targeted Amino-Acid Composition
Basten-Romero, V., Muñoz-Tafalla, R., Díaz-Rovira, A. M., Miquel-Oliver, B., Filella-Merce, I., & Guallar, V. (2026) · arXiv.org · doi:10.48550/arxiv.2606.27939 · Cited by 1
“…which uses conditioning tags the model was trained to recognize (Madani et al., 2023; Munsamy et al., 2024), to supervised fine-tuning (FT) (Madani et al., 2023) and, more recently, reward-guided FT via reinforcement learning (RL) (Cao et al., 2025; Stocco et al., 2024; Subramanian et al.,…”
InfluentialHuang, Y., Ge, R., Wu, L., Sun, Y., Wang, Z., & Cao, M. (2026) · ACS Synthetic Biology · doi:10.1021/acssynbio.6c00199 · Cited by 1
Reconstructability of evolutionary intermediates in generative epistatic landscapes
Netti, R., & Weigt, M. (2026) · bioRxiv · doi:10.64898/2026.06.18.733143 · Cited by 1
“The general ideas are, however, directly applicable to other generative modeling frameworks, including deep-learning architectures, in particular when formulated in an autoregressive way (which we use naturally for conditioning a model for S mid to S start and S end ) [6, 8, 9, 41].”
Zhang, S., Han, Y., Ding, Y., Lin, X., Yuan, Z., Sun, X., Yuan, Q., Liu, Z., & Zhang, Y. (2026) · Nature Communications · doi:10.1038/s41467-026-74506-x · Cited by 3
“Notably, deep learning models have shown success in generating candidate amino acid sequences tailored to given protein backbones or desired functions 32,33 .”
ADAPT: a programme for the advanced detection of AI-enabled pathogenic threats
Pálya, H., & Nelson, C. (2026) · Frontiers in Bioengineering and Biotechnology · doi:10.3389/fbioe.2026.1819372 · Cited by 1
Reddy, S. T. (2026) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.64898/2026.06.08.730844
“It follows the autoregressive language-model paradigm developed for biological-sequence generation, which has been applied to protein sequences in ProGen 36 and to SMILES in MolGPT 37 ; what is new here is the cross-attention conditioning on the dtSFM encoder's target embedding, which yields target-specific molecular…”
Photobiological Hydrogen Production in Cyanobacteria: Advances, Challenges, and Perspectives
Tang, W., Cheng, Z.-H., Li, W., Li, P.-S., Chen, M., & Fu, Y. (2026) · Fermentation · doi:10.3390/fermentation12060273 · Cited by 1
“In addition, AI-guided protein redesign has been successfully applied to generate functional protein sequences with improved stability and catalytic properties, highlighting the potential of data-driven approaches for rational hydrogenase engineering [111].”
Ahmad, M., Iqbal, M. W., Mirza, M. U., Raza, M. A., Alghamdi, H. A., Haider, S. Z., Nawaz, M. Z., & Yu, K. (2026) · Enzyme and Microbial Technology · doi:10.1016/j.enzmictec.2026.110927 · Cited by 1
Programming biology: next-gen AI firms raise billions to design better medicines
Senior, M. R. (2026) · Nature Biotechnology · doi:10.1038/s41587-026-03170-9 · Cited by 1
Fernández, A. T., Fernández, A., Méndez-Líter, J. A., Prieto, A., Barriuso, J., & Osorio, F. (2026) · bioRxiv · doi:10.64898/2026.05.20.726739 · Cited by 1
“The use of PLMs for enzyme optimization has been previously reported 14,26,42 , yet their predictive accuracy in this domain has generally been lower compared to applications such as PLM-driven antibody optimization 43,44 .”
Weber, J. K., Parajuli, G., Wang, S. H., Ratner, V., Ma, X., Shoshan, Y., Zhang, L., Morrone, J. A., Raboh, M., Hexter, E., et al. (2026) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.64898/2026.05.15.725176 · Cited by 1
“To approach this problem, recent advancements in generative AI, LLMs, and diiusion-based artificial intelligence (AI) architectures have been used to design protein-based therapeutics ( 2, 9-20 ).”
CatIF-RL: Activity-Oriented Enzyme Sequence Design by Steered Inverse Protein FoldingPreprint
Li, Y.-H., Xiong, J.-L., Zhang, Y.-X., Cai, T., Fu, C., Li, S.-T., Xu, W., Lyu, R., Chen, Z.-Y., Guo, Z., et al. (2026) · bioRxiv · doi:10.64898/2026.05.11.724288 · Cited by 1
“This observation suggests that the pipeline may be steering the policy toward sequence regions that are sparsely populated by natural enzymes but remain structurally plausible under the surrogate ensemble, consistent with guided exploration of dark sequence space 19,20 .”
Pravin Kumar, R., Sigamani, G., Umesh, V. N., Lalitha, R., Natarajan, D., Kumar, M., & Karuppasamy, R. (2026) · ACS Sustainable Chemistry & Engineering · doi:10.1021/acssuschemeng.5c10179 · Cited by 1
Spinner, A., Notin, P., Berry, S. P., Cortade, D. L., Sisson, Z., Ikonomova, S. P., Ross, D. L., & Marks, D. S. (2026) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.64898/2026.05.08.723784 · Cited by 1
“As a result, machine learning approaches have been introduced to predict the effects of mutations and suggest new variants with improved or altered function (Madani & al., 2023; Shin et al., 2021; Lian et al., 2022; Schiff et al., 2024; Sum-ida et al., 2024; Thadani et al., 2023).”
Sanders, J., Giancardo, L., Guo, L., Zhao, Y., Sönmez, K., Cheng, N., & Yilmaz, M. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2605.06720 · Cited by 1
“ProGen [22] and ProGen2 [27] condition an autoregressive model on functional family annotations to enable controllable generation across broad protein families.”
Retrieval and competition: how a protein foundation model starts a protein
Jedryszek, P., & Crook, O. (2026) · arXiv.org · doi:10.48550/arxiv.2605.16331 · Cited by 1
“From predicting mutational effects (Meier et al., 2021) to annotating protein function (Chen et al., 2025) and guiding the design of novel proteins (Madani et al., 2023), protein language models (PLMs) achieve strong performance despite being trained only on sequence data, suggesting that they…”
Deep learning-driven integrated pipeline for de novo design and synthesis of antimicrobial peptides
Liu, J., Chen, Y., Tang, J., Xing, X., Lin, J., Sun, J., Xing, X., Li, J., & Zhang, C. Y. (2026) · npj Drug Discovery. · doi:10.1038/s44386-026-00045-6 · Cited by 3
“Models such as GPT 11 and LLaMA 12 in natural language processing, alongside ProtGPT2 13 and ProGen 14 in the domain of protein design, had demonstrated exceptional capabilities in capturing the properties of biological macromolecules through pre-training on extensive protein sequence corpora.”
Virtual cell: Current perspectives and future prospects
Li, J., Zhang, H., Zhang, W., Li, N., Zhao, F., & Shen, G. (2026) · Journal of International Medical Research · doi:10.1177/03000605261425080 · Cited by 1
“74–76 Although we have endeavored to provide a comprehensive roadmap of the VC landscape, certain limitations of this review should be noted.”
Generative design of intrinsically disordered protein regions with IDiom
Liu, J., Ibarraran, S., Hu, F., Park, A., Dunn, A. R., & Rotskoff, G. M. (2026) · bioRxiv · doi:10.64898/2026.04.10.717777 · Cited by 1
“These models learn rich evolutionary statistics over amino acid sequences and have enabled the design of novel proteins and functional variants [15, 16, 17, 18, 19, 20].”
Li, X., Luo, X., Xie, S., Wei, Y., Hong, F., Jiang, Y., Zhong, X., Xie, X., Ma, C., Hao, Y., et al. (2026) · Current Research in Food Science · doi:10.1016/j.crfs.2026.101404 · Cited by 1
A Generative Neuro‐Symbolic AI for Protein Sequence DesignPreprint
Defresne, M., Dessaux, D., Buchet, S., Barthe, L., Ammar-Khodja, L., Azizi, B., Durante, V., Cioci, G., de Givry, S., Roussel, A., et al. (2026) · bioRxiv · doi:10.64898/2026.03.31.715526 · Cited by 1
Tang, Z., Wang, S., Jiang, L., Li, D., & Zhu, G. (2026) · Journal of environmental chemical engineering · doi:10.1016/j.jece.2026.122523
Selvaraj, A., Jayawant, M., Kayalvizhi, N., Krishnan, M., Puvaneshvari, N., Kumar, A. W. S., & Rameshkumar, N. (2026) · Methods · doi:10.1016/j.ymeth.2026.04.006 · Cited by 1
DMAPLM: A multimodal pretrained framework for computational drug repositioning
Chen, H., & Li, Z. (2026) · PLoS Computational Biology · doi:10.1371/journal.pcbi.1014192 · Cited by 1
“Pretrained models, such as BioBERT [15], ChemBERTa [16], and ESM-2 [17], have shown strong feature representation capabilities by learning from large-scale corpora, and successfully applied to tasks such as molecular property prediction [18] and protein design [19].”
Lin, H., Vogel, H., & Zhang, H. (2026) · International Journal of Molecular Sciences · doi:10.3390/ijms27073142 · Cited by 5
CAPSUL: A Comprehensive Human Protein Benchmark for Subcellular Localization
Hu, Y., Lin, X.-Y., Li, S., Wang, W.-J., Zhu, F., & Feng, F.-L. (2026) · arXiv.org · doi:10.48550/arxiv.2603.18571 · Cited by 1
“Subsequently, Transformer-based models have demonstrated strong performance, especially after large-scale pretraining, achieving impressive results across a range of downstream tasks (Rives et al., 2019; Lin et al., 2022; Madani et al., 2023).”
Ferraz, L., Rodrigues, A. F., Cotovio, P. G., Ventura, M., Silva, G., Coroadinha, A. S., Machuqueiro, M., & Pesquita, C. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2603.19473
RMSAGen: Integrating Multiple Sequence Alignment for Function RNA Design
Jiang, J.-Y., Chen, Y.-Y., Zhang, Q., Li, J., Shi, X.-Y., Zhou, C., Lin, Z.-Q., Wang, J.-M., He, D.-C., Hong, L., et al. (2026) · AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v40i1.37012 · Cited by 1
“Transformers (Vaswani et al. 2017) demonstrate outstanding textual understanding and generating capabilities even on non-natural languages such as biological molecular sequences (Ji et al. 2021; Ferruz, Schmidt, and H¨ocker 2022b; Madani et al. 2023; Jiang et al. 2025a; Wang et al. 2025).”
ST-LLM: Spatial Transcriptomics Embedding with Large Language Models
Xu, Z.-T., Wan, X.-H., Li, L., Feng, S., Zhang, Y.-M., Zhang, F., & Hu, B. (2026) · AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v40i40.40713 · Cited by 1
“LLMs have demonstrated revolutionary application potential in bioinformatics, particularly in single-cell transcriptomics analysis (Lin et al. 2023; Rives et al. 2021; Madani et al. 2023).”
Generalise or Memorise? Benchmarking Ligand-Conditioned Protein Generation from Sequence-Only DataPreprint
Vicente-Sola, A., Dornfeld, L., Coines, J., & Ferruz, N. (2026) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.64898/2026.02.06.704305
“These advances have enabled the design of diverse proteins and enzymes, including lysozymes(Madani et al., 2023), triose-phosphate isomerases(Romero-Romero et al., 2024), malate dehydrogenases(Johnson et al., 2025), or nucleases(Ivanˇci´c et al., 2025).”
Data storage and retrieval with unnatural proteins expressed via E. coli
Zhou, Y., Ng, C. C. A., Liu, C., Tam, W. M., Lau, F. C. M., & Yao, Z. (2026) · Nature Communications · doi:10.1038/s41467-026-70061-7 · Cited by 1
Lipinski, C., Tanifuji, R., Connolly, J. A., Breitling, R., Oguri, H., & Takano, E. (2026) · Current Opinion in Biotechnology · doi:10.1016/j.copbio.2026.103454 · Cited by 1
AI-driven discovery and engineering of human-derived protein scaffolds for mRNA delivery
Xun, D., Li, L., Zou, J., Liu, Y., Yang, Z., Liang, Q., Li, K., Zhu, Z., Zhang, J., Lang, L., et al. (2026) · Nature Communications · doi:10.64898/2026.02.07.704552
Design prokaryotic cis-regulatory elements using language model
Xia, Y., Sun, J., Du, X., Liang, Z., Wu, X., Shi, W., Shao, B., Guo, S., & Huo, Y. (2026) · Nucleic Acids Research · doi:10.1093/nar/gkag122 · Cited by 2
“From a machine learning perspective, design promoters for various species are different tasks, and in most cases, the task-specific training data are limited [ 29 ].”
Protein Language Visualizer: A Repository for Homology Exploration with Language Model EmbeddingsPreprint
Espinoza-Herrera, J., Manríquez-García, M. F., Medina-Bermejo, S., López-Jasso, A., Ruiz-Alcocer, J. P., Siordia, A., Veskimägi, S. M., Roethler, N., & Jinich, A. (2026) · bioRxiv · doi:10.1101/2024.11.19.624229 · Cited by 1
“Built on transformer architectures with attention and positional encoding [23, 24], they are trained through masked-sequence prediction to learn context-dependent representations of amino acids.”
The application of generative neural radiance fields in medical imaging: A systematic review
Kabika, T. B., Cai, H.-S., Said, R., Dong, J., Zimba, B., Pinda, O., Zhang, S.-Y., Li, Y., & Hou, W.-G. (2026) · Medical Physics (Lancaster) · doi:10.1002/mp.70296 · Cited by 1
“5 Within healthcare, these approaches serve distinct purposes: predicting protein structures from sequences 6,7 and performing medical image tasks such as segmentation (labeling anatomy or pathology) and reconstruction (inferring volumetric or higher-quality images from limited data).”
AI Developments for T and B Cell Receptor Modeling and Therapeutic Design
Xie, L., Pélissier, A., Shao, Y.-J., & Martínez, M. (2026) · doi:10.48550/arxiv.2601.17138 · Cited by 1
“This enables open-ended sequence generation without predefined length constraints, making them well-suited for scaffold-free design (for example, ProGen and related AR models [68]).”
Accelerating inference in genomic and proteomic foundation models via speculative decoding
Provatas, K., Karatzikos, A., Koilakos, C., Patsakis, M., Tzanakakis, A., Nayak, A., Mouratidis, I., Avgoulas, E. I., & Georgakopoulos-Soares, I. (2026) · bioRxiv · doi:10.64898/2026.01.13.699044 · Cited by 1
“The same principles of predictive parallelism and multi-stage verification are highly relevant for computational biology, where generation of several candidate genomes or proteins is often required to achieve optimal results (Merchant et al. 2025; Madani et al. 2023).”
Meng, Q. (2026) · arXiv.org · doi:10.48550/arxiv.2601.00647 · Cited by 1
“Among unaligned backbones, ProGen2-XL provides the strongest baseline with 52.4% fold-ability, yet nearly half of the sequences remain non-foldable, indicating persistent structural hallucinations.”
InfluentialMachine learning reveals hidden dimensions of functional similarity in proteins
Youssef, N., Gurev, S., & Marks, D. S. (2026) · Proceedings of the National Academy of Sciences of the United States of America · doi:10.1073/pnas.2524802122 · Cited by 1
Unraveling circadian rhythms—computational insights into molecular mechanisms
Rao, Y., & Srivastava, A. (2026) · FEBS Letters · doi:10.1002/1873-3468.70264
“Another study developed ProGen, a protein language model that is trained on evolutionary sequence data and can generate artificial proteins with novel functions [142].”
Application and progress of artificial intelligence agents in drug development
Zhao, D., & Hsieh, C. (2026) · Journal of Zhejiang University (Medical Sciences) · doi:10.3724/zdxbyxb-2025-0697 · Cited by 1
Ko, Y. S., Parkinson, J., & Wang, W. (2026) · Briefings in Bioinformatics · doi:10.1093/bib/bbag014 · Cited by 1
“Protein language models (pLMs) such as ESM [1] have catalyzed progress across a wide range of p rotein modeling tasks, including variant effect prediction [2], structure prediction [1], and protein engineering [3, 4].”
Deep Learning Meets Synthetic Biology: Applications Across Molecular, Cellular, and Metabolic Scales
Zhao, Y., Yang, C., Xue, B., Liu, H., Wang, S., & Su, H. (2026) · Green Carbon · doi:10.1016/j.greenca.2025.11.003 · Cited by 2
Benchmarking Chemical, Genetic, and Cell Line Encodings for Cancer Perturbation Response PredictionPreprint
Zinchenko, V., Schlicker, A., Kurilov, R., Pouplin, A., Villalba, S., & Horlacher, M. (2025) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.64898/2025.12.15.694331
“Protein language models (pLMs) are ubiquitously used to predict 3D structure and functional properties of proteins, score the impact of missense variants or design new proteins with desired propreties (21, 59–62).”
De novo generation of functional terpene synthases using TpsGPTPreprint
Ramanathan, H., Bushuiev, R., Soldát, M., Kohout, J., Hebra, T., Smith, J. D., Šivic, J., & Pluskal, T. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2512.08772 · Cited by 1
“Additionally, the fine-tuning is typically done on large models such as ProGEN with 280 million parameters [Madani et al., 2023].”
Learning RNA language through pre-trained models: From sequence to structure
Zhang, S., & Chen, S.-J. (2025) · Biophysics Reviews · doi:10.1063/5.0273433 · Cited by 1
Gabrielli, J., Lewis, N. E., Kontoravdi, C., & Ceroni, F. (2025) · Cell Systems · doi:10.1016/j.cels.2025.101424 · Cited by 1
Unveiling the landscape of prokaryotic global regulators through deep protein language models
Geng, J., Wu, J., Luo, S., Liu, D., Nie, J., Fan, G., Sun, Q., Hu, S., & Wu, L. (2025) · mSystems · doi:10.1128/msystems.00950-25 · Cited by 1
He, Z.-C., Bian, M.-X., Zhu, J., Chen, J.-Y., Wang, Y., Zhao, W., Li, T.-B., Han, B., Xu, J., & Wu, J. (2025) · arXiv.org · doi:10.48550/arxiv.2511.13107 · Cited by 1
“From assisting in clinical documentation and drafting manuscripts to accelerating drug discovery, LLMs are poised to reshape the research landscape [21][22][23].”
Boosting In-Silicon Directed Evolution with Fine-Tuned Protein Language Model and Tree SearchPreprint
Yang, Y., Wang, Y., Li, J., Pei, G., Han, D., Chen, G., & Heng, P. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2511.09900 · Cited by 1
“…al., 2019), and a lot of in-silicon directed evolution algorithms (Brookes & Listgarten, 2018; Sinai et al., 2020; Ren et al., 2022; Wang et al., 2023) have been proposed to strategically explore the protein fitness landscape to identify desired sequence mutations via an iterative search process.”
Accessible biocatalyst development by rapid in vitro semi-rational engineering (RISE) of enzymes
Telek, A., Gaják, V. Z., Dargó, G., Mičan, J., Bednář, D., Vértessy, B. G., & Tasnádi, G. (2025) · iScience · doi:10.1016/j.isci.2025.114257 · Cited by 1
“15,16 A major advantage of CFPS is its ability to express proteins from linear template DNA.”
Protein language model supervised motif-scaffolding design with GPDL
Zhang, B., Liu, K., Zheng, Z., Zhu, J., Li, Z., Liu, Y., Mu, J., Wei, T., & Chen, H. (2025) · International Journal of Biological Macromolecules · doi:10.1016/j.ijbiomac.2025.148441 · Cited by 1
Hu, Q., Tian, X., Li, Y., Zhou, R., Wang, Z., Meng, J., Wang, S., Guo, J., Li, W., Zheng, L., et al. (2025) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2025.07.19.665668 · Cited by 1
Explainable machine learning reveals evolutionary signals in Influenza hemagglutinin
Meyer, A. (2025) · bioRxiv · doi:10.1101/2025.09.21.677610 · Cited by 1
“While LightGBM is a powerful algorithm, other architectures, including deep learning models like transformers that have shown promise in protein modeling [33, 35], may capture different and important aspects of sequence evolution.”
Property-Isometric Variational Autoencoders for Sequence Modeling and Design
Sadeghi, E., Deng, X., Lin, I.-H., Copp, S. M., & Bogdanov, P. (2025) · arXiv.org · doi:10.48550/arxiv.2509.14287 · Cited by 1
“More recently transformer architectures, pre-trained on large protein or RNA corpora or fine-tuned on SELEX datasets, have emerged as reliable tools for aptamer design and family-agnostic protein generation [25–27].”
Data-optimal scaling of paired antibody language modelsPreprint
Neyestanak, M. S., Burbach, S. M., Ng, K., Gangavarapu, P., Hurtado, J., Magura, J., Ismail, N., Muema, D., Ndung’u, T., Ward, A. B., et al. (2025) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2025.09.02.673765 · Cited by 1
Large language models for biological sequence analysis in infectious disease research
Luo, J., Cai, X., & Li, Y. (2025) · Biosafety and Health · doi:10.1016/j.bsheal.2025.09.007 · Cited by 7
An Artificial Intelligence Model for Translating Natural Language into Functional de Novo ProteinsPreprint
Riley, T. P., Parsa, M. S., Kalantari, P., Naderi, I., Azimian, K., Begloo, N., Matusovsky, O., Osher, E. L., Prabhakaran, D., Yadav, R., et al. (2025) · bioRxiv · doi:10.1101/2025.03.21.644400 · Cited by 1
“…ATPases extends the small but growing canon of ML-designed enzymes—including pGAN59 (malate dehydrogenase; ProteinGAN), FastPETase (poly(ethylene terephthalate) hydrolysis; MutCompute), L056 (lysozyme; ProGen), and momi120_102 (serine hydrolase; PLACER = RFdiffusion+LigandMPNN) ( 11 , 46 – 48 ).”
Provasi, D., Konovalov, K., Riina, N., Cullen, O., & Filizola, M. (2025) · bioRxiv · doi:10.1101/2025.08.04.668547 · Cited by 1
“…enhance these deep learning architectures 12 and accurately predict ligand bioactivities for GPCRs with limited data, we recently explored the synergy between these models and large language models (LLMs) designed for protein modeling 16, 17 in predicting GPCR low-efficacy and biased agonists.”
Denger, A., & Helms, V. (2025) · Molecules · doi:10.3390/molecules30153226 · Cited by 1
“This includes feature extraction from protein sequences [23,24], protein sequence generation from text [25–27], as well as various predictive tasks, e.g., for secondary and tertiary structure [28–30], structural similarity and protein family membership [31,32], GO term annotations [33–37],…”
In silico design of smaller size enzymatic protein by generative artificial intelligence (ProtGPT2)
Hamada, H., Matsuzawa, T., & Hanai, T. (2025) · Journal of Bioscience and Bioengineering · doi:10.1016/j.jbiosc.2025.06.009 · Cited by 1
Frank, R., Widrich, M., Akbar, R., Klambauer, G., Sandve, G. K., Robert, P. A., & Greiff, V. (2026) · PLoS Computational Biology · doi:10.1371/journal.pcbi.1014805 · Cited by 1
Ren, C.-Y., He, D., & Huang, J. (2025) · doi:10.48550/arxiv.2503.21123 · Cited by 1
“• ProGen (Madani et al., 2023): ProGen is a language model leveraging a state-of-the-art Transformer architecture.”
Large Scale Multi-Task Bayesian Optimization with Large Language Models
Zeng, Y.-M., Maus, N., Jones, H., Tao, J., Wan, F.-P., Torres, M. D. T., de la Fuente-Nunez, C., Marcus, R., Bastani, O., & Gardner, J. R. (2025) · arXiv.org · doi:10.48550/arxiv.2503.08131 · Cited by 1
“This set of techniques has been applied to biophysical domains such as molecular design and protein engineering, where the LLM proposes mutations to enhance certain properties, as well as to program optimization tasks where the LLM speeds up code execution time [36, 45, 21].”
Generalizable Metamaterials Design Techniques Inspire Efficient Mycelial Materials Inverse Design
Zavorskas, J., Edwards, H., Marten, M. R., Harris, S. D., & Srivastava, R. (2025) · ACS Biomaterials Science & Engineering · doi:10.1021/acsbiomaterials.4c01986 · Cited by 2
Al-Shouha, M., & Szűcs, G. (2025) · Multimedia Systems · doi:10.1007/s00530-024-01629-w · Cited by 1
“They are utilized in a wide range of applications from writing software [1] to generating functional protein sequences [2].”
Open-Source Protein Language Models for Function Prediction and Protein Design
Pandi, S. V., & Ramsundar, B. (2024) · arXiv.org · doi:10.48550/arxiv.2412.13519 · Cited by 1
“L G ] 18 D ec 2024 In protein design, PLMs are proving to be powerful tools for generating novel proteins with tailored properties, a task typically requiring substantial experimental effort(Madani et al. 2020)(Madani et al. 2023).”
Gómez‐Pérez, D., & Keller, A. (2025) · NAR Genomics and Bioinformatics · doi:10.1093/nargab/lqaf174
“Recent progress in NLP, including the development and high popularity of large language models (LLMs) based on transformer architecture, has opened new avenues in biological research [ 13 , 14 ].”
The path to adoption of open source AI for drug discovery in Africa
Turón, G., & Duran‐Frigola, M. (2024) · Artificial Intelligence in the Life Sciences · doi:10.1016/j.ailsci.2024.100118 · Cited by 1
Yang, E.-W., & Velazquez-Villarreal, E. (2024) · medRxiv · doi:10.1101/2024.11.27.24318113 · Cited by 1
Walking by design: how to build artificial molecular motors made of proteins
Nilsson, P. F., Zink, A., Laprévote, O. M. C., Korosec, C. S., Gustafsson, N., Robertson, N. O., Forde, N. R., Eichhorn, R., Höcker, B., Curmi, P. M. G., et al. (2024) · Nano Futures · doi:10.1088/2399-1984/ad7d7e · Cited by 2
Gene synthesis design: a pythonic approach
Hu, Y.-Z., Pan, D., Xu, F., Huang, B.-F., Chen, X.-Y., & Lin, S.-Q. (2024) · PeerJ · doi:10.7717/peerj.17750 · Cited by 1
Porokhin, V., Brown, A. M., & Hassoun, S. (2025) · Journal of Computer-Aided Molecular Design · doi:10.1007/s10822-025-00613-4 · Cited by 1
“Further, emerging methods for molecular [54, 55] and protein generation [56, 57] enable the creation of enhanced structures using ML techniques.”
Assessing the laboratory performance of AI-generated enzymes
Zelezniak, A., Yang, K. K., & Johnson, S. J. (2024) · Nature Biotechnology · doi:10.1038/s41587-024-02239-7 · Cited by 1
AbDPP: Target‐oriented antibody design with pretraining and prior biological structure knowledge
Yu, C., Lin, X., Cheng, Y., Xu, J., Wang, H., Yan, Y., Huang, Y., Liu, L., Zhao, W., Zhao, Q., et al. (2024) · Proteins Structure Function and Bioinformatics · doi:10.1002/prot.26676 · Cited by 1
“Although current deep learning approaches for antibody sequence generation show promise, 20,23,25 there remains considerable room for improvement.”
Metadata
- Authors
- Madani, A., Krause, B., Greene, E. R., Subramanian, S., Mohr, B. P., Holton, J. M., Olmos, J. L., Xiong, C., Sun, Z. Z., Socher, R., Fraser, J. S., & Naik, N.
- Affiliations
- Salesforce (United States) (US)
- University of California, San Francisco (US)
- Howard Hughes Medical Institute (US)
- Lawrence Berkeley National Laboratory (US)
- Corresponding author
- Madani, Ali
- Naik, Nikhil
- ORCID
- Venue
- Nature Biotechnology
- Published
- 2023-01-26
- Volume
- 41
- Issue
- 8
- Pages
- 1099-1106
- ISSN
- 0733-222X (ISSN-L)
- 1087-0156
- 1546-1696
- PMC
- PMC10400306
- OpenAlex
- W4318071656
- Cited by
- 1,078
- References
- 80
- Citations by year
- 2026: 314 · 2025: 320 · 2024: 302 · 2023: 141
- FWCI
- 95.97
- Type
- journal article
- Language
- en
- Open access
- green
- www.ncbi.nlm.nih.gov
- Licence
- Locations
- Nature Biotechnology (publishedVersion)
- PubMed (publishedVersion)
- PubMed Central (acceptedVersion)
- eScholarship (California Digital Library) (submittedVersion)
- OSTI OAI (U.S. Department of Energy Office of Scientific and Technical Information) (submittedVersion)
- Funders
- U.S. Department of Health & Human Services | National Institutes of Health (GM123159)
- Topics
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Protein Structure and Dynamics
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Machine Learning in Bioinformatics
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Enzyme Catalysis and Immobilization
- Keywords
- large language models, protein sequence generation, protein design, protein families, lysozyme, chorismate mutase, malate dehydrogenase, catalytic efficiency, controllable generation, protein engineering
- MeSH terms
- Amino Acid Sequence, Chorismate Mutase, Estrogens, Conjugated (USP), Language, Proteins
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
A language model drew the claims from the paper on 2026-10-05, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-05 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
