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 401–500 of 1,012
Predictive and therapeutic applications of protein language models
Furui, K., Sakano, K., & Ohue, M. (2025) · Allergology International · doi:10.1016/j.alit.2025.08.004 · Cited by 12
MD-LLM-1: A Large Language Model for Molecular DynamicsPreprint
Murtada, M. H., Brotzakis, Z. F., & Vendruscolo, M. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2508.03709 · Cited by 9
“LMs have already been exploited in a range of applications in protein science 37 , including for protein structure prediction 38 and protein design 39 .”
Li, Z., Chen, Y., Chen, X., Guo, Z., Guan, G., Feng, Y., & Chen, H. (2025) · World Journal of Microbiology and Biotechnology · doi:10.1007/s11274-025-04475-8 · Cited by 11
“Madani et al. (2023 ) developed the Pro-Gen language model, which generates functional protein sequences across large families with predictable activities.”
PFMBench: Protein Foundation Model Benchmark
Gao, Z.-Y., Wang, H., Tan, C., Xu, C., Liu, M.-D., Hu, B.-Z., Chao, L.-L., Zhang, X.-M., & Li, S. (2025) · arXiv.org · doi:10.48550/arxiv.2506.14796 · Cited by 10
“Since the debut of ESM-1B [53] in 2021, a diverse array of PFMs—spanning various architectures and training paradigms—has emerged [53, 35, 18, 11, 41, 12, 59, 80, 10, 5, 68, 55, 56, 71, 4, 16, 33].”
mCLM: A Modular Chemical Language Model that Generates Functional and Makeable Molecules
Edwards, C. N., Han, C., Lee, G., Nguyen, T., Szymkuć, S., Prasad, C. K., Jin, B., Han, J.-W., Diao, Y., Liu, G., et al. (2025) · International Conference on Learning Representations · doi:10.48550/arxiv.2505.12565 · Cited by 10
“Protein language models (Lin et al., 2023; Hayes et al., 2025; Wang et al., 2025b; Madani et al., 2023; Xiao et al., 2024b; Su et al., 2024; Buehler and Buehler, 2024; Wang et al., 2025a; Zhu et al., 2024; Ferruz et al., 2022) have been highly successful because proteins/peptides are tok-enized at…”
CellVerse: Do Large Language Models Really Understand Cell Biology?
Zhang, F., Liu, T., Zhu, Z.-H., Wu, H., Wang, H., Zhou, D.-H., Zheng, Y.-F., Wang, K., Wu, X., & Heng, P.-A. (2025) · Neural Information Processing Systems · doi:10.48550/arxiv.2505.07865 · Cited by 10
“Recent advances in large language models (LLMs) have spurred interest in applying them to scientific domains, including mathematics [24, 4], chemistry [47, 59], and biology [40, 42].”
Bayless, A., Song, L.-J., Sorbello, M., Ogden, S. C., Todd, T., Flint, A., Maruta, N., Tulu, J., Drenichev, M., Ntoukakis, V., et al. (2025) · bioRxiv · doi:10.1073/pnas.2505893122 · Cited by 10
Zhu, X., Qin, C., Wang, F., Yang, F., He, B., Zhao, Y., & Yao, J. (2024) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2024.06.24.600337 · Cited by 8
“Lysozyme Generation For the lysozyme generation task, we leveraged a dataset sourced from Progen [68].”
AlignAb: Pareto-Optimal Energy Alignment for Designing Nature-Like Antibodies
Wen, Y.-B., Xu, C.-W., Hu, J. Y.-C., Ding, K.-Z., & Liu, H. (2024) · Neural Information Processing Systems · doi:10.48550/arxiv.2412.20984 · Cited by 10
“For example, Madani et al. [2023] and Rives et al. [2019] borrow ideas from language models and treat proteins as sequences to predict their structures, functions, and other important properties.”
Chen, A., Stanton, S., Ding, F., Alberstein, R. G., Watkins, A. M., Bonneau, R., Gligorijević, V., Cho, K., & Frey, N. C. (2024) · arXiv (Cornell University) · doi:10.48550/arxiv.2410.22296 · Cited by 7
“One common approach is to use control codes , or unique strings pre-pended in front of a training example that indicates which attribute is represented in the example’s target (Keskar et al., 2019; Padmakumar et al., 2023; Raffel et al., 2020; Madani et al., 2023).”
Learning Biophysical Dynamics with Protein Language ModelsPreprint
Hou, C., Zhao, H.-Q., & Shen, Y. (2024) · bioRxiv · doi:10.1101/2024.10.11.617911 · Cited by 10
“Recently, protein language models (pLMs) like ESM1,2 7, 8 , ProtTrans 9 , and ProGen 10 have emerged as powerful tools for generating implicit EBR.”
Function-Guided Conditional Generation Using Protein Language Models with AdaptersPreprint
Yang, J., Bhatnagar, A., Ruffolo, J. A., & Madani, A. (2024) · arXiv (Cornell University) · doi:10.48550/arxiv.2410.03634 · Cited by 11
“…sequences with a target function, such as specific enzymatic activity, involves finetuning a PLM on a specific class or family of enzymes and then generating novel enzymes with that function (Madani et al., 2023; Ruffolo et al., 2024b) – an approach we will refer to as Target Finetuning .”
Self-citationGenoTEX: An LLM Agent Benchmark for Automated Gene Expression Data AnalysisPreprint
Liu, H., Chen, S., Zhang, Y., & Wang, H. (2024) · arXiv (Cornell University) · doi:10.48550/arxiv.2406.15341 · Cited by 19
Autophagy and machine learning: Unanswered questions
Yang, Y., Pan, Z., Sun, J., Welch, J. D., & Klionsky, D. J. (2024) · Biochimica et Biophysica Acta (BBA) - Molecular Basis of Disease · doi:10.1016/j.bbadis.2024.167263 · Cited by 10
ProteinEngine: Empower LLM with Domain Knowledge for Protein Engineering
Shen, Y., Lv, O., Zhu, H.-Y., & Wang, Y.-G. (2024) · Conference on Artificial Intelligence in Medicine in Europe · doi:10.48550/arxiv.2405.06658 · Cited by 10
“Although LLMs have been explored for specific tasks within the protein engineering domain such as protein structure prediction [11], protein evolution analysis, or de novo protein design [6,13], their application often demands significant alterations to the model architecture.”
Cytometry masked autoencoder: An accurate and interpretable automated immunophenotyper
Kim, J., Ionita, M., Lee, M., McKeague, M. L., Pattekar, A., Painter, M. M., Wagenaar, J., Truong, V., Norton, D. T., Mathew, D., et al. (2024) · Cell Reports Medicine · doi:10.1016/j.xcrm.2024.101808 · Cited by 13
“However, the revolution of self-supervised learning techniques, particularly in the pre-training phase, empowers models to more accurately learn data distributions and utilize unlabeled data effectively.”
Data-driven molecular design and simulation in modern chemical engineering
Gartner, T. E., Ferguson, A. L., & Debenedetti, P. (2024) · Nature Chemical Engineering · doi:10.1038/s44286-023-00010-4 · Cited by 10
Domrös-Zoungrana, D., Rajaeean, N., Boie, S. D., Fröling, E., & Lenz, C. (2024) · Journal of Medical Education and Curricular Development · doi:10.1177/23821205241284719 · Cited by 19
Strategies and mechanisms of precision genome engineering: From gene editing to genome writing
Huang, K., Tian, J., Zhao, W., Zhang, S., Xiang, L., Yang, M., Ls, L., Huang, Y., Zhang, N., Sun, L., et al. (2026) · iMetaOmics. · doi:10.1002/imo2.70115 · Cited by 8
Yang, Z., & Tang, T. (2026) · Biotechnology Advances · doi:10.1016/j.biotechadv.2026.108908 · Cited by 6
Dual-encoder contrastive learning accelerates enzyme discovery
Rocks, J. W., Truong, D. P., Rappoport, D., Maddrell-Mander, S., Martin-Alarcon, D. A., Lee, T. M., Crossan, S., & Goldford, J. E. (2026) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.2520070123 · Cited by 9
“Recent advances in protein language models demonstrate that scaling to larger and more diverse datasets consistently improves performance (19–21).”
AI-driven antimicrobial peptide characterization unveils novel motifs for drug design
Padi, S., Mondal, K., Hoogerheide, D., Heinrich, F., Mihailescu, M., Klauda, J. B., & Cardone, A. (2025) · Scientific Reports · doi:10.1038/s41598-025-30419-1 · Cited by 9
“Moreover, the peak at k-mer lengths 14, and 18 indicates strong reliability for motifs associated with beta sheets.”
Artificial intelligence in protein-based detection and inhibition of AMR pathways
Sadhukhan, S., Bhattacharya, R., Bhattcharya, D., Sahana, S., Pradhan, B., Pandit, S., Gill, H. S., Rajeev, M., Nag, M., & Lahiri, D. (2025) · Journal of Computer-Aided Molecular Design · doi:10.1007/s10822-025-00710-4 · Cited by 10
A vision–language pretrained transformer for versatile clinical respiratory disease applications
Ma, L., Liang, H., He, Y., Wang, W., Yan, Z., Li, W., Wang, R., Li, Y., Lizhu, Y., Liu, Y., et al. (2025) · Nature Biomedical Engineering · doi:10.1038/s41551-025-01544-z · Cited by 9
Pseudodata-based molecular structure generator to reveal unknown chemicals
Yu, N., Ma, Z., Shao, Q., Li, L., Wang, X. Y., Pan, B., Yu, H., & Wei, S. (2025) · Nature Machine Intelligence · doi:10.1038/s42256-025-01140-5 · Cited by 8
“Our MSGo is trained with tens of thousands of virtual mass spectra using a transformer neural network previously applied in protein prediction 28 .”
Bioengineering approaches to trained immunity: Physiologic targets and therapeutic strategies
Knight, H. R., Kim, M., Kannan, N., Taylor, H. B., Main, H., Azcue, E., & Esser-Kahn, A. P. (2025) · eLife · doi:10.7554/elife.106339 · Cited by 9
“…and ProGen could be utilized to design a specialized protein-based therapy, such as a constitutively active metabolic enzyme that produces metabolic intermediates, like mevalonate or itaconate, which are known to induce training ( Madani et al., 2023 ; Guan et al., 2025 ; Luo et al., 2024 ).”
ProtChatGPT: Towards Understanding Proteins with Hybrid Representation and Large Language Models
Wang, C., Fan, H., Quan, R., Yao, L., & Yang, Y. (2025) · International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR) · doi:10.1145/3726302.3730064 · Cited by 9
“Considering protein sequences as language, several works [29, 34, 36, 51] encode amino acid tokens using transformer model [47] to extract pairwise relationships among amino acids, and autoregressively recover protein sequences based on extensive protein sequence databases.”
Toward the Explainability of Protein Language Models
Hunklinger, A., & Ferruz, N. (2025) · doi:10.48550/arxiv.2506.19532 · Cited by 9
“…DNA sequencing and the extension of protein structure databases have enabled the training of ever-larger protein language models (pLMs). pLMs now set the state of the art in tasks as diverse as the prediction of drug-target interactions 2 , protein structure prediction 3,4 , and protein design 5 .”
Designing Cyclic Peptides via Harmonic SDE with Atom-Bond Modeling
Zhou, X., Li, M.-Y., Xiao, Y., Li, J.-H., Xue, D., Zheng, Z., Ma, J.-Z., & Gu, Q. (2025) · International Conference on Machine Learning · doi:10.48550/arxiv.2505.21452 · Cited by 9
“ProGen (Madani et al., 2023) shows an autoregressive language model trained on massive protein sequence data can utilize property tags for controllable sequence generation.”
Large Language Models for EDA: Future or Mirage?
He, Z., Pu, Y., Wu, H., Qiu, T., & Yu, B. (2025) · ACM Transactions on Design Automation of Electronic Systems · doi:10.1145/3736167 · Cited by 21
Integrating experimental feedback improves generative models for biological sequences
Calvanese, F., Peinetti, G., Pavlinova, P., Nghe, P., & Weigt, M. (2025) · bioRxiv · doi:10.1101/2025.03.31.646327 · Cited by 9
“However, a persistent challenge remains: these models often produce a high rate of false positives – sequences generated as potentially functional by the model but failing in experimental tests.”
Wang, B., & Li, W. (2024) · Genes · doi:10.3390/genes15081090 · Cited by 7
“Besides protein structure prediction, there is evidence that pLMs also perform well in various other predictive modeling tasks, including protein function annotation [39,40], protein design [41,42], and ligand binding prediction [43,44].”
BackgroundRapid protein evolution by few-shot learning with a protein language modelPreprint
Jiang, K.-Y., Yan, Z.-Q., Bernardo, M. D., Sgrizzi, S. R., Villiger, L., Kayabolen, A., Kim, B., Carscadden, J. K., Hiraizumi, M., Nishimasu, H., et al. (2024) · bioRxiv · doi:10.1101/2024.07.17.604015 · Cited by 9
“Generative PLMs, such as ESM3, ProtGPT2, and ProGen( 7 – 9 ), can design novel proteins, but these de novo-designed variants typically only reach wild-type level activity after extensive rounds of experimental testing( 7 , 10 ).”
Foundation models in molecular biology
Si, Y., Zou, J., Gao, Y., Chuai, G., Liu, Q., & Chen, L. (2024) · Biophysics Reports · doi:10.52601/bpr.2024.240006 · Cited by 6
Floating Anchor Diffusion Model for Multi-motif Scaffolding
Liu, K., Mao, W.-A., Shen, S., Jiao, X.-R., Sun, Z., Cheng, H., & Shen, C.-H. (2024) · International Conference on Machine Learning · doi:10.48550/arxiv.2406.03141 · Cited by 9
“The advent of generative protein models (De Bortoli et al., 2022; Lee et al., 2023; Madani et al., 2023; Trippe et al., 2023; Gruver et al., 2023; Lisanza et al., 2023) has instigated a dramatic evolution in protein design.”
BackgroundSelf-citationBaseNet: A transformer-based toolkit for nanopore sequencing signal decoding
Li, Q.-W., Sun, C., Wang, D., & Lou, J. (2024) · bioRxiv · doi:10.1016/j.csbj.2024.09.016 · Cited by 9
Learning Beyond Pattern Matching? Assaying Mathematical Understanding in LLMsPreprint
Guo, S., Didolkar, A., Ke, N. R., Goyal, A., Huszár, F., & Schölkopf, B. (2024) · arXiv (Cornell University) · doi:10.48550/arxiv.2405.15485 · Cited by 6
“With the promising success, there is a growing trend to use LLM as creative assistants for scientific discovery: from mathematics (Trinh et al., 2024) to biology (Madani et al., 2023).”
BackgroundDescriptor-augmented machine learning for enzyme-chemical interaction predictions
Han, Y., Zhang, H., Zeng, Z., Liu, Z., Lu, D., & Liu, Z. (2024) · Synthetic and Systems Biotechnology · doi:10.1016/j.synbio.2024.02.006 · Cited by 11
“The capacity of deep learning algorithms to capture higher-order correlations in protein sequences may explain why they could outperform those artificially constructed descriptors [47,48].”
BackgroundEmpirical validation of ProteinMPNN’s efficiency in enhancing protein fitness
Wang, T., Jin, X., Lu, X., Min, X., Ge, S., & Li, S. (2024) · Frontiers in Genetics · doi:10.3389/fgene.2023.1347667 · Cited by 9
Patterns Lead the Way to Far-from-Equilibrium Materials
Knoll, P., Ouyang, B., & Steinbock, O. (2023) · ACS Physical Chemistry Au · doi:10.1021/acsphyschemau.3c00050 · Cited by 9
Protein language models uncover carbohydrate-active enzyme function in metagenomics
Thurimella, K., Mohamed, A. M. T., Li, C., Vatanen, T., Graham, D. B., Owens, R. M., La Rosa, S. L., Plichta, D. R., Bacallado, S., & Xavier, R. J. (2025) · BMC Bioinformatics · doi:10.1186/s12859-025-06286-y · Cited by 9
Schmidinger, N., Schneckenreiter, L., Seidl, P., Schimunek, J., Hoedt, P.-J., Brandstetter, J., Mayr, A., Luukkonen, S., Hochreiter, S., & Klambauer, G. (2024) · arXiv · doi:10.48550/arxiv.2411.04165 · Cited by 8
“In summary, Prot-xLSTM outperforms larger unconditional protein language models like ESM-2 (Lin et al., 2023) and ProGen2 (Nijkamp et al., 2023), and matches or surpasses ProtMamba’s performance.”
Modeling patient tissues at molecular resolution with EvaPreprint
Liu, Y., Sharma, R., Bieniosek, M. F., Kang, A., Wu, E., Chou, P., Li, I., Rahim, M. K., Bauer, E., Ji, R., et al. (2025) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.64898/2025.12.10.693553 · Cited by 3
“In biology, foundation model training frameworks have been established for several different types of data, including protein [1,2], DNA [3] and RNA sequences [4], single-cell transcriptomes [5], digital pathology [6–8], and microscopy images [9].”
Chien, K. R., Foo, K. S., & Witman, N. (2026) · Nature Reviews Drug Discovery · doi:10.1038/s41573-026-01459-7 · Cited by 12
Towards the explainability of protein language models
Hunklinger, A., & Ferruz, N. (2026) · Nature Machine Intelligence · doi:10.1038/s42256-026-01232-w · Cited by 12
“…of DNA sequencing and the extension of protein structure databases have enabled the training of ever-larger protein language models (pLMs). pLMs now set the state of the art in tasks as diverse as the prediction of drug–target interactions 2 , protein structure prediction 3,4 and protein design 5 .”
Technologies to measure and modulate protein subcellular localization
Leineweber, W. D., Tei, R., Mäkiniemi, A., Ting, A. Y., & Lundberg, E. K. (2026) · Nature Reviews Molecular Cell Biology · doi:10.1038/s41580-026-00957-1 · Cited by 8
A deep learning and large language hybrid workflow for omics interpretation
Tang, D., Zhang, C., Zhang, W., Lu, F., Xiao, L., Huang, X., Shao, J., Liu, D., Fu, S., Zhao, M., et al. (2026) · Nature Biomedical Engineering · doi:10.1038/s41551-025-01576-5 · Cited by 5
“In addition to natural language comprehension and generation, other transformer-based LLMs, including ESM 78 , ProGen 79 , scBERT 80 , Gene-former 81 and scGPT 82 , have been developed for processing downstream tasks across various biological scenarios.”
A PINN-based transformer model for heat transfer prediction in freeze-drying process
Cheng, Y., Chen, H., Zhou, X., Ye, H., & Pan, T. (2026) · International Communications in Heat and Mass Transfer · doi:10.1016/j.icheatmasstransfer.2025.109985 · Cited by 8
GeoEvoBuilder: A deep learning framework for efficient functional and thermostable protein design
Liu, J., You, H., Guo, Z., Xu, Q., Zhang, C., & Lai, L. (2025) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.2504117122 · Cited by 7
AI-guided Cas9 engineering provides an effective strategy to enhance base editing
Wei, D., Cheng, P., Song, Z., Liu, Y., Xu, X., Huang, X., Wang, X., Zhang, Y., Shu, W., & Wei, Y. (2025) · Molecular Systems Biology · doi:10.1038/s44320-025-00142-0 · Cited by 6
“Meanwhile, advances in arti fi cial intelligence technology have opened new opportunities for protein engineering (Abramson et al, 2024; AlQuraishi, 2021; Jumper et al, 2021; Madani et al, 2023; Notin et al, 2024; Riesselman et al, 2018; Senior et al, 2020; Yeh et al, 2023).”
Biomarker integration and biosensor technologies enabling AI-driven insights into biological aging
Kushner, J. A., Pandey, M., & Kohli, S. S. S. (2025) · Frontiers in Aging · doi:10.3389/fragi.2025.1703698 · Cited by 7
“Recent work has already demonstrated the feasibility of using language-based generative models to design protein structures that bind previously undruggable targets (Madani et al., 2023).”
CRISPR meets AlphaFold: guiding SWEET10-enhanced oil production.
Li, L.-J., Zhang, Z.-Y., & Zhang, B. (2025) · Trends in Plant Science · doi:10.1016/j.tplants.2025.08.001 · Cited by 8
Chen, J. Z., Gall, B., Pulsford, S. B., Tokuriki, N., & Jackson, C. J. (2025) · Molecular biology and evolution · doi:10.1093/molbev/msaf136 · Cited by 8
“…the traditional homology-based neighborhoods in SSNs by using sequence vector representations from protein language models (pLM) (Alley et al. 2019; Madani et al. 2023; Rives et al. 2021) for sequence space exploration, and find them to perform comparably to the homology-based method in…”
Adapting ProteinMPNN for antibody design without retrainingPreprint
del Alamo, D., Frick, R., Truan, D., & Karpiak, J. D. (2025) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2025.05.09.653228 · Cited by 7
Prediction of Umami Peptides Based on a Large Language Model of Proteins
He, Y., Tian, Z., Zheng, J., Wang, H., Han, L., & Han, W. (2025) · Journal of Chemical Information and Modeling · doi:10.1021/acs.jcim.4c02394 · Cited by 8
Yang, W., Wang, Y., & Wang, Y. (2026) · Biophysics Reviews · doi:10.1063/5.0273394 · Cited by 9
Miniaturizing, Modifying, and Magnifying Nature’s Proteins with RaygunPreprint
Devkota, K., Shonai, D., Mao, J., Ko, Y. S., Wang, W., Soderling, S., & Singh, R. (2025) · bioRxiv · doi:10.1101/2024.08.13.607858 · Cited by 8
“…conservation to learn dis-22 tributional rules, generating rich, high-dimensional embeddings of protein sequences: a sequence of n amino acids maps to n points in a high-dimensional space, with each point encapsulating both local and global context of its corresponding residue [24, 27, 21, 6, 14].”
Mapping the evolution of generative AI: insights from bibliometric research
Pathak, S., & Pallasena, R. K. (2025) · Journal of Decision System · doi:10.1080/12460125.2024.2438612 · Cited by 11
Song, J., & Kurgan, L. (2025) · Expert Review of Proteomics · doi:10.1080/14789450.2025.2451715 · Cited by 9
“In particular, we observe a recent trend in the development of PLMs that have begun to target specific classes/ families of proteins, with examples of ProGen that focuses on certain families of lysozymes [95] and IgLM on antibodies [96].”
Yang, S., Ni, J., & Xu, P. (2024) · Journal of Agricultural and Food Chemistry · doi:10.1021/acs.jafc.4c05650 · Cited by 8
Chen, Z., Chen, T., Xie, C.-G., Xue, Y., Zhang, X.-N., Zhou, J.-B., & Fang, X. (2024) · arXiv.org · doi:10.48550/arxiv.2407.09274 · Cited by 8
“…aims to generate sequences that can fold into specific three-dimensional structures (structure-to-sequence) [10, 3, 11] and to create protein sequences [12, 13, 14, 15] or structures [15] based on a given tag [12, 13] or description [14, 15] (description-to-sequence or description-to-structure).”
BackgroundTan, Y., Zheng, L., Zhong, B., Hong, L., & Zhou, B. (2024) · IEEE International Conference on Bioinformatics and Biomedicine · doi:10.1109/bibm62325.2024.10822035 · Cited by 8
FusOn-pLM: A Fusion Oncoprotein-Specific Language Model via Focused Probabilistic MaskingPreprint
Vincoff, S., Goel, S., Kholina, K., Pulugurta, R., Vure, P., & Chatterjee, P. (2024) · bioRxiv · doi:10.1101/2024.06.03.597245 · Cited by 8
“However, these models were not trained on fusion oncoprotein sequences, which are functionally and structurally distinct from their wild-type counterparts due to their altered binding sites and unique breakpoint junctions.”
RhoMax: Computational Prediction of Rhodopsin Absorption Maxima Using Geometric Deep Learning
Sela, M., Church, J. R., Schapiro, I., & Schneidman‐Duhovny, D. (2024) · Journal of Chemical Information and Modeling · doi:10.1021/acs.jcim.4c00467 · Cited by 7
PandoGen: Generating complete instances of future SARS-CoV-2 sequences using Deep Learning
Ramachandran, A., Lumetta, S., & Chen, D.-M. (2023) · bioRxiv · doi:10.1101/2023.05.10.540124 · Cited by 8
“…of proteins which contains sequences belonging to various different protein families sourced from various different organisms such as those found in the UniProt database [uni17], and (2) they are finetuned on the specific subfamily of proteins that are desired to be generated [MKG + 23] [NRW + 22].”
InfluentialBAGEL: Protein engineering via exploration of an energy landscape
(2025) · doi:10.1371/journal.pcbi.1013774 · Cited by 7
“Regarding EmbeddingOracle s, we plan to explore models including AMPLIFY [48] and ProtGen [49], as well as protein language models trained explicitly to provide improved descriptions of protein-protein interactions such as the recently released MINT [50].”
AI-enabled alkaline-resistant evolution of protein to apply in mass production
Kang, L., Wu, B., Zhou, B., Tan, P., Kang, Y. (., Yan, Y., Zong, Y., Li, S., Liu, Z., & Hong, L. (2025) · eLife · doi:10.7554/elife.102788 · Cited by 7
“Recent advances in large-scale protein language models (LLMs) have enabled zero-shot predictions of protein mutants based on self-supervised learning from natural protein sequences ( Hie et al., 2021 ; Madani et al., 2023 ; Rao et al., 2021 ; Rives et al., 2021 ).”
Functional protein design and enhancement with ontology reinforcement iteration
He, B., Qin, C., Zhao, Y. S., Huang, L.-K., Wu, Z., Wang, F., Wu, F., Yang, F., & Yao, J. (2026) · Nature Communications · doi:10.1038/s41467-026-69855-6 · Cited by 7
Koehl, A., Prillo, S., Liu, M., Xiong, J. T., Weng, L., Savage, D. F., & Song, Y. S. (2026) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.64898/2026.02.19.706898 · Cited by 5
Lamb, K. D., Hughes, J., Lytras, S., Young, F., Koci, O., Herzig, J. C., Lovell, S. C., Grove, J., Yuan, K., & Robertson, D. L. (2026) · Nature Communications · doi:10.1038/s41467-026-69569-9 · Cited by 20
“They are helping improve the effectiveness of antibodies 56 and even generate new proteins 57 .”
Decode-gLM: Tools to Interpret, Audit, and Steer Genomic Language ModelsPreprint
Maiwald, A., Jedryszek, P., Draye, F., Schölkopf, B., Morris, G. M., & Crook, O. M. (2025) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2025.10.31.685860 · Cited by 6
“This progress mirrors breakthroughs using protein language models (Ferruz et al., 2022; Lin et al., 2023; Nijkamp et al., 2023), capable of designing novel, functional fluorescent proteins (Hayes et al., 2025) or lysozymes (Madani et al., 2023).”
Eskandar, K. (2025) · AI and Ethics · doi:10.1007/s43681-025-00872-9 · Cited by 11
“Protein language models trained on evolutionary datasets capture multi-scale structure–function signals and allow zero-shot predictions of mutational effects, accelerating prioritization of variants (Rives et al. 2021 ; Madani et al. 2023 ).”
Li, X., Chen, W., Wang, L., Zheng, P., Xu, J.-B., Tao, X., Yu, H., & Xu, S.-C. (2025) · Journal of Hazardous Materials · doi:10.1016/j.jhazmat.2025.140581 · Cited by 7
Large language models for Alzheimer's disease drug discovery
Alkam, T., Tarshizi, E., & Benschoten, A. H. V. (2025) · Journal of Alzheimer s Disease · doi:10.1177/13872877251346890 · Cited by 9
“30 It uses deep learning to generate arti fi cial protein sequences that are functional across diverse protein families.”
Enhanced prediction of antigen and antibody binding interface using ESM-2 and Bi-LSTM
Li, Q., Zhao, Y., Chordia, M. D., Xia, X., Zhang, B., & Zheng, H. (2025) · Human Immunology · doi:10.1016/j.humimm.2025.111304 · Cited by 7
Roex, G., Gordon, K. S., Lion, E., Birnbaum, M. E., & Anguille, S. (2025) · Journal for ImmunoTherapy of Cancer · doi:10.1136/jitc-2024-010658 · Cited by 7
“102 For a comprehensive discussion on how artificial intelligence can advance cellular therapies, including CAR- T-cell therapy, we refer to a review by Caponni and Daniels.”
NTxPred2: A large language model for predicting neurotoxic peptides and neurotoxins
Rathore, A., Jain, S., Choudhury, S., & Raghava, G. (2025) · bioRxiv · doi:10.1101/2025.03.01.640936 · Cited by 7
“These protein embeddings are crucial for various applications, including structure prediction, novel sequence generation, and protein classification (Bepler & Berger, 2021; Madani et al., 2023).”
Fine-tuning of conditional Transformers improves in silico enzyme prediction and generation
Nicolini, M., Saitto, E., Franco, R. E. J., Cavalleri, E., Galeano, A., Malchiodi, D., Paccanaro, A., Robinson, P. N., Casiraghi, E., & Valentini, G. (2025) · Computational and Structural Biotechnology Journal · doi:10.1016/j.csbj.2025.03.037 · Cited by 8
“Our recent results on the fine-tuning of the ProGen model on the lysozyme family of proteins achieved a statistically signifi-cant improvement in terms of accuracy and perplexity [26], confirming previous results on a similar task [14].”
InfluentialModern machine learning methods for protein property prediction.
Dosajh, A., Agrawal, P., Chatterjee, P., & Priyakumar, U. (2025) · Current Opinion in Structural Biology · doi:10.1016/j.sbi.2025.102990 · Cited by 7
Reliable and Diverse Evaluation of LLM Medical Knowledge Mastery
Zhou, Y., Liu, X., Ning, C., Zhang, X., & Wu, J. (2024) · International Conference on Learning Representations · doi:10.48550/arxiv.2409.14302 · Cited by 7
“Recent years have witnessed the rapid advancement of large language models (LLMs), which have exhibited potential across various domains (Brown et al., 2020; Ouyang et al., 2022; Touvron et al., 2023; OpenAI, 2023; Madani et al., 2023; Boiko et al., 2023), including medicine.”
BackgroundLamb, K. D., Hughes, J., Lytras, S., Young, F., Koci, O., Herzig, J. C., Lovell, S., Grove, J., Yuan, K., & Robertson, D. L. (2024) · bioRxiv · doi:10.1101/2024.07.05.602129 · Cited by 7
“They are helping improve the effectiveness of antibodies 45 and even generate new proteins 46 .”
Aggarwal, N., Rathore, M., Jan, F., Sharma, D., Kumar, S., Thudi, M., Jighly, A., Varshney, R. K., & Mir, R. R. (2026) · Theoretical and Applied Genetics · doi:10.1007/s00122-026-05293-8 · Cited by 8
“Through extensive data mining, the study team created the most thorough CRISPR ‒ Cas Atlas (Madani et al. 2023) available.”
Liang, Y., & Wang, C. (2026) · Journal of Chemical Information and Modeling · doi:10.1021/acs.jcim.5c02745 · Cited by 5
EvoFlows: Evolutionary Edit-Based Flow-Matching for Protein Engineering
Deutschmann, N., Ferragu, C., Ziegler, J. D., Aziznejad, S., & Bixby, E. (2026) · arXiv.org · doi:10.48550/arxiv.2603.11703 · Cited by 6
“Autoregressive models have been successful for de novo -like tasks such as enzyme generation with weak conditioning (Madani et al., 2023; Munsamy et al., 2024).”
Controlling Repetition in Protein Language ModelsPreprint
Zhang, J., Zhang, Z., Wang, D., & Hu, L. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2602.00782 · Cited by 5
“They have enabled impressive advances in structure prediction and de novo protein design (Lin et al., 2022; Hayes et al., 2025; Ferruz et al., 2022; Madani et al., 2023; Li et al., 2024; Hesslow et al., 2022).”
Tang, J., Gong, D., Li, H., & Li, S. (2026) · Acta Pharmaceutica Sinica B · doi:10.1016/j.apsb.2026.01.039 · Cited by 6
“By conditioning the generation process on functional annotations ( e.g. , enzyme commission numbers), ProGen can create novel, biologically plausible protein sequences with specific, desired functions 81 .”
Rong, D.-Y., Chen, Z.-J., Jia, Q., Zhang, K.-W., Lu, H., Zhai, G.-T., & Liu, N. (2025) · doi:10.48550/arxiv.2512.22257 · Cited by 6
“Large Language Models (LLMs) have recently become a new paradigm for advancing protein research Madani et al. (2023); Xiao et al. (2025b).”
Ko, Y. S., Parkinson, J., & Wang, W. (2025) · bioRxiv · doi:10.1101/2024.08.24.609531 · Cited by 6
“Protein language models (pLMs) such as ESM [1] have catalyzed progress across a wide range of protein modeling tasks, including variant effect prediction [2], structure prediction [1], and protein engineering [3,4].”
Before LUCA: unearthing the chemical roots of metabolism
Carbonell, P., & Peretó, J. G. (2025) · Philosophical Transactions of the Royal Society B Biological Sciences · doi:10.1098/rstb.2024.0292 · Cited by 8
Wang, Y.-K., Li, M., Huang, F., Qiu, M.-Y., & Zhang, W. (2025) · Advancement of science · doi:10.1002/advs.202507457 · Cited by 6
Microalgal bioengineering for futuristic applications in synthetic and space biology
Shelake, R. M., Khalid, M. A., & Kim, J. (2025) · Biotechnology Advances · doi:10.1016/j.biotechadv.2025.108665 · Cited by 14
HARPS: A Hybrid Algorithm for Robust Plant Stress Detection to Foster Sustainable Agriculture
Hussain, S. M., Jeong, B., Mir, B. A., & Lee, S. (2025) · Sustainability · doi:10.3390/su17135767 · Cited by 6
Large-scale protein clustering in the age of deep learning
Pereira, J., Pantolini, L., Durairaj, J., & Schwede, T. F. (2025) · Current Opinion in Structural Biology · doi:10.1016/j.sbi.2025.103078 · Cited by 7
Precise genome editing process and its applications in plants driven by AI
Jiang, B., An, Z., Niu, L., & Qin, D. (2025) · Functional & Integrative Genomics · doi:10.1007/s10142-025-01619-9 · Cited by 9
Tremmel, R., Honoré, A., Park, Y., Zhou, Y.-T., Xiao, M., & Lauschke, V. (2025) · Pharmacogenomics (London) · doi:10.1080/14622416.2025.2504863 · Cited by 6
Tan, Y., Liu, C., Gao, J.-Y., Wu, B., Li, M.-C., Wang, R.-L., Zhang, L., Yu, H.-Q., Fan, G.-S., Hong, L., et al. (2025) · arXiv.org · doi:10.48550/arxiv.2503.15438 · Cited by 6
“…data and predictive modules, these models facilitate downstream tasks such as sequence generation, catalytic activity enhancement, function prediction, and properties assessment, thereby advancing enzyme production and application (Madani et al., 2023; Zhou et al., 2024b; Kang et al., 2024).”
T-cell receptor specificity landscape revealed through de novo peptide designPreprint
Visani, G. M., Pun, M. N., Minervina, A. A., Bradley, P., Thomas, P., & Nourmohammad, A. (2025) · bioRxiv · doi:10.1101/2025.02.28.640903 · Cited by 6
“The PAE values reported by the structure prediction algorithms have been shown to be noisy indicators of protein folding fidelity and design quality [21].”
Protein evolution as a complex system
Gall, B., Pulsford, S. B., Matthews, D. S., Spence, M. A., Kaczmarski, J. A., Chen, J. Z., Sandhu, M., Stone, E. A., Nichols, J., & Jackson, C. J. (2025) · Nature Chemical Biology · doi:10.1038/s41589-025-01977-2 · Cited by 7
“Identifying and understanding the interactions of these forces is central to describing how this system Check for updates BOX 1”
Shuvo, M. H., & Bhattacharya, D. (2024) · Computational and Structural Biotechnology Journal · doi:10.1016/j.csbj.2024.12.015 · Cited by 5
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
