The DOI matches its Crossref record; 130 of the 133 references it cites were found in a registry or an index; none of the 53 sources its claims cite is checked for support yet.
Learning the protein language: Evolution, structure, and function
Open accessAbstract
Language models have recently emerged as a powerful machine-learning approach for distilling information from massive protein sequence databases. From readily available sequence data alone, these models discover evolutionary, structural, and functional organization across protein space. Using language models, we can encode amino-acid sequences into distributed vector representations that capture their structural and functional properties, as well as evaluate the evolutionary fitness of sequence variants. We discuss recent advances in protein language modeling and their applications to downstream protein property prediction problems. We then consider how these models can be enriched with prior biological knowledge and introduce an approach for encoding protein structural knowledge into the learned representations. The knowledge distilled by these models allows us to improve downstream function prediction through transfer learning. Deep protein language models are revolutionizing protein biology. They suggest new ways to approach protein and therapeutic design. However, further developments are needed to encode strong biological priors into protein language models and to increase their accessibility to the broader community.
Overview
AI snapshot
Written from the paper’s full text.
TL;DR
Protein language models learn evolutionary, structural, and functional organization from large sequence databases, and adding structure supervision improves downstream prediction through transfer learning. The authors present a multi-task LSTM trained on 76 million sequences plus 28,000 structures and show gains in structural similarity, transmembrane prediction, and sequence-to-phenotype prediction.
Key findings
- A multi-task LSTM (MT-LSTM) trained with masked language modeling, residue-residue contact prediction, and SCOP structural similarity prediction learned embeddings that “dramatically outperform” sequence comparison methods and even outperform TMalign for structural similarity; structure supervision was critical for fine-grained organization, and MT-LSTM outperformed the prior SSA-LSTM.
- For transmembrane region prediction, BiLSTM+CRF models using MT-LSTM embeddings significantly outperformed DLM-LSTM embeddings (paired t-test, p=0.044) and also outperformed existing predictors and SSA-LSTM.
- For sequence-to-phenotype prediction on deep mutational scanning datasets, MT-LSTM embeddings with Gaussian process regression achieved an average Spearman correlation of 0.65, significantly outperforming ECNet’s 0.60 (paired t-test, p=0.006).
- The paper notes that language models predicted SARS-CoV-2 spike mutations with high antibody-escape potential, but as of publication viral escape had not yet been experimentally verified.
Methods
- The authors trained a three-layered bidirectional LSTM with skip connections and 1024 hidden units in each direction per layer. The MT-LSTM combined three losses: masked language modeling (10% of amino acids randomly masked), residue-residue contact prediction (contacts defined as residues within 8 Å), and SCOP structural similarity prediction via ordinal regression; MT-LSTM loss weights were λmasked=0.5, λcontact=0.9, λsimilarity=0.1, while DLM-LSTM used only λmasked=1.
- Training data were 76,215,872 UniRef90 sequences (July 2018) and 28,010 SCOPe ASTRAL 2.06 protein sequences with structures, split into 22,408 training and 5,602 testing sequences. Models were trained for 1,000,000 parameter updates with Adam at learning rate 0.0001 on a single NVIDIA V100 GPU; training took roughly 13 days for DLM-LSTM and 51 days for MT-LSTM.
- Evaluations included structural similarity on 100,000 randomly sampled SCOPe ASTRAL test pairs, transmembrane prediction on the TOPCONS2 dataset of 5,154 proteins using 10-fold cross validation, and sequence-to-phenotype prediction on deep mutational scanning datasets using 5-fold cross validation and Gaussian process regression with an RBF kernel.
Limitations
- The paper states that further developments are needed to encode strong biological priors into protein language models and to increase their accessibility to the broader community; it also states that current models make heavily simplified phylogenetic assumptions, treating sequences as independent draws from a star phylogeny.
- The paper states practical limitations including high compute costs, murky design guidelines, ad hoc model development, quadratic scaling of transformers with sequence length (requiring long sequences to be excluded or truncated), and poor learning of long-range dependencies by RNNs.
- Evident from the text: the new results come from a single MT-LSTM architecture and specific benchmarks, and the SARS-CoV-2 escape predictions had not been experimentally verified as of publication.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
0 of 53 cited sources checked · 0 supported · 0 flagged
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 find that increasing model size improves transfer learning performance.”
§ Learning the Protein Language: Evolution, Structure and Function > Transfer learning improves downstream applications
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesA 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
“However, this organization is improved when we include structure supervision in the language model training (Figure 4b).”
§ Learning the Protein Language: Evolution, Structure and Function > Multi-task language models capture the semantic organization of proteins
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesA 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
“Furthermore, representations learned by our MT-LSTM model significantly outperform (paired t-test, p=0.044) the embeddings learned by our DLM-LSTM model on this application (Figure 5b).”
§ Learning the Protein Language: Evolution, Structure and Function > Transfer learning improves downstream applications
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIn the paper
“Furthermore, the multi-task learning approach outperforms a two-step learning approach presented previously (SSA-LSTM)27.”
§ Learning the Protein Language: Evolution, Structure and Function > Multi-task language models capture the semantic organization of proteins
What the sources say
- Not checked yetcites [27] Bepler T & Berger B Learning protein sequence embeddings using information from structure. in International Conference on Learning Representations (2019). PMC12947628 41768824 source
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesA 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
“Because we embed sequences into a semantic representation space, we can find structurally related proteins even though their sequences are not closely related (Figure 4c, Supplemental Table 1).”
§ Learning the Protein Language: Evolution, Structure and Function > Multi-task language models capture the semantic organization of proteins
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesA 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 observe that even though the DLM-LSTM model was trained using only sequence information, protein sequences still organize roughly by structure in embedding space.”
§ Learning the Protein Language: Evolution, Structure and Function > Multi-task language models capture the semantic organization of proteins
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesA 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
“Because TMscore is not symmetric, we calculate TMscore for both comparison directions and average them together for each protein pair. We found this outperformed other methods of combining the two scores.”
§ Learning the Protein Language: Evolution, Structure and Function > Methods > Protein structural similarity prediction evaluation
Strength:plain· not yet calibratedCausal:direct· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes? Load-bearing: unclear✓ Based: passesIn the paper
“The key insight was that the sequence representations (vector embeddings) learned by the language model were powerful features for solving other prediction problems.”
§ Learning the Protein Language: Evolution, Structure and Function > Transfer learning improves downstream applications
What the sources say
- Not checked yetcites [27] Bepler T & Berger B Learning protein sequence embeddings using information from structure. in International Conference on Learning Representations (2019). PMC12947628 41768824 source
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIn the paper
“Interestingly, because so many sequences are available, these models continue to benefit from increased size 32.”
§ Learning the Protein Language: Evolution, Structure and Function > Protein language models distill information from massive protein sequence databases
What the sources say
- Not checked yetcites [32] Rives A Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences. bioRxiv (2019). PMC8053943 33876751 source
Strength:plain· not yet calibratedCausal:direct· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes? Load-bearing: unclear✓ Based: passesIn the paper
“Learned protein representations improve predictive ability for downstream prediction problems through transfer learning (Figure 5a).”
§ Learning the Protein Language: Evolution, Structure and Function > Transfer learning improves downstream applications
What the sources say
- Not checked yetcites [27] Bepler T & Berger B Learning protein sequence embeddings using information from structure. in International Conference on Learning Representations (2019). PMC12947628 41768824 source
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passes- Rests on an assumption: Many problems in biology are small data problems; quantitative protein characterization assays are rarely high throughput and methods are needed that can generalize given only 10s to 100s of experimental measurements.
- Rests on an assumption: We are often interested in extrapolating from data collected over a small region of protein sequence space to other sequences, often with little to no homology.
- It may state two separate claims as one. 13
- The quote only partly states this claim. 1
- Probably not a substantial claim. 8
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 find that our learned semantic embeddings dramatically outperform the sequence comparison methods and even outperform structure comparison with TMalign when predicting structural similarity.”
§ Learning the Protein Language: Evolution, Structure and Function > Multi-task language models capture the semantic organization of proteins
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt 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 knowledge distilled by these models allows us to improve downstream function prediction through transfer learning.”
- Also stated in: § Multi-task language models capture the semantic organization of proteins · boosted
“Here, we will demonstrate that training protein language models with self-supervision on a large amount of natural sequence data and with structure supervision on a smaller set of sequence, structure pairs enriches the learned representations and translates into improvements in downstream prediction problems (Figure 3).”
- Also stated in: § Transfer learning improves downstream applications · plain
“Learned protein representations improve predictive ability for downstream prediction problems through transfer learning (Figure 5a).”
§ Learning the Protein Language: Evolution, Structure and Function > Abstract
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
- Rests on an assumption: Many problems in biology are small data problems; quantitative protein characterization assays are rarely high throughput and methods are needed that can generalize given only 10s to 100s of experimental measurements.
- Rests on an assumption: We are often interested in extrapolating from data collected over a small region of protein sequence space to other sequences, often with little to no homology.
- Also stated in: § Multi-task language models capture the semantic organization of proteins · boosted
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 find that we can predict the phenotypes of unobserved sequence variants across datasets better than existing methods (Figure 5c).”
§ Learning the Protein Language: Evolution, Structure and Function > Transfer learning improves downstream applications
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one. The quote only partly states this claim.
- Rests on an assumption: Many problems in biology are small data problems; quantitative protein characterization assays are rarely high throughput and methods are needed that can generalize given only 10s to 100s of experimental measurements.
- Rests on an assumption: We are often interested in extrapolating from data collected over a small region of protein sequence space to other sequences, often with little to no homology.
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 find that the BiLSTM+CRFs with our new embeddings (DLM-LSTM and MT-LSTM) outperform existing transmembrane predictors and a BiLSTM+CRF using our previous smaller embedding model (SSA-LSTM).”
§ Learning the Protein Language: Evolution, Structure and Function > Transfer learning improves downstream applications
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt 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
“Interestingly, we observe that the structural supervision component is critical for learning well organized embeddings at a fine-grained level, because the DLM-LSTM representations alone do not perform well at this task (Supplemental Table 1).”
§ Learning the Protein Language: Evolution, Structure and Function > Multi-task language models capture the semantic organization of proteins
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
In the paper
“This parallels the general trend in natural language processing, where the number of parameters, rather than specific architectural choices, is the best indicator of model performance 52.”
§ Learning the Protein Language: Evolution, Structure and Function > Protein language models distill information from massive protein sequence databases
What the sources say
- Not checked yetcites [52] Kaplan J Scaling Laws for Neural Language Models. arXiv [cs.LG] (2020).
Strength:plain· not yet calibratedCausal:correlational· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✗ Load-bearing: fails✓ Based: passesProbably not a substantial claim. The paper’s conclusions may not depend on it.
None of the sources cited for this claim have been checked yet.
In the paper
“Although structure is more informative of function than sequence 57,58 and structure is encoded by sequence, predicting structure remains hard, particularly due to the relative paucity of structural relative to sequence data.”
§ Learning the Protein Language: Evolution, Structure and Function > Supervision encodes biological meaning
What the sources say
- Not checked yetcites [57] Zhang C & Kim S-H Overview of structural genomics: from structure to function. Curr. Opin. Chem. Biol 7, 28– 32 (2003). 12547423 source
- Not checked yetcites [58] Shin DH Structure-based inference of molecular functions of proteins of unknown function from Berkeley Structural Genomics Center. J. Struct. Funct. Genomics 8, 99– 105 (2007). 17764033 source
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
None of the sources cited for this claim have been checked yet.
In the paper
“They also learn summary representations, powerful features that can be used to better capture sequence relationships and link sequence to function via transfer learning 27,28,32-35.”
§ Learning the Protein Language: Evolution, Structure and Function > Introduction
What the sources say
- Not checked yetcites [27] Bepler T & Berger B Learning protein sequence embeddings using information from structure. in International Conference on Learning Representations (2019). PMC12947628 41768824 source
- Not checked yetcites [28] Alley EC, Khimulya G, Biswas S, AlQuraishi M & Church GM Unified rational protein engineering with sequence-based deep representation learning. Nat. Methods 16, 1315– 1322 (2019). 31636460 PMC7067682 source
- Not checked yetcites [32] Rives A Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences. bioRxiv (2019). PMC8053943 33876751 source
- Not checked yetcites [33] Rao R Evaluating Protein Transfer Learning with TAPE. Adv. Neural Inf. Process. Syst 32, 9689– 9701 (2019). 33390682 PMC7774645 source
- Not matchedcites [34] Luo Y Evolutionary Context-Integrated Deep Sequence Modeling for Protein Engineering. in Research in Computational Molecular Biology 261– 263 (Springer International Publishing, 2020).
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- Not checked yetcites [35] Hie B, Bryson BD & Berger B Leveraging Uncertainty in Machine Learning Accelerates Biological Discovery and Design. Cell Syst 11, 461– 477.e9 (2020). 33065027 source
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
None of the sources cited for this claim have been checked yet.
In the paper
“Although it is generally believed that proteins with similar sequences form similar structures, there are also interesting examples of highly similar protein sequences having radically different structures and functions 60,61 and of sequences that can form multiple folds 62.”
§ Learning the Protein Language: Evolution, Structure and Function > Supervision encodes biological meaning
What the sources say
- Not checked yetcites [60] Kosloff M & Kolodny R Sequence-similar, structure-dissimilar protein pairs in the PDB. Proteins 71, 891– 902 (2008). 18004789 PMC2673347 source
- Not checked yetcites [61] Wei KY Computational design of closely related proteins that adopt two well-defined but structurally divergent folds. Proc. Natl. Acad. Sci. U. S. A 117, 7208– 7215 (2020). 32188784 PMC7132107 source
- Not checked yetcites [62] James LC & Tawfik DS Conformational diversity and protein evolution--a 60-year-old hypothesis revisited. Trends Biochem. Sci 28, 361– 368 (2003). 12878003 source
Strength:plain· not yet calibratedCausal:correlational· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes? Load-bearing: unclear✓ Based: passesIt 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
“For example, unrelated protein sequences can form the same structural fold and, therefore, are semantically similar. However, we cannot deduce this relationship from sequences alone. Supervision is required to learn that these sequences belong to the same semantic category.”
§ Learning the Protein Language: Evolution, Structure and Function > Supervision encodes biological meaning
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt 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 phenomena suggest that there are aspects of protein biology that may not be discoverable by statistical sequence models alone. Supervision that represents known protein structure, function, and other prior knowledge may be necessary to encode distant sequence relationships into learned embeddings.”
§ Learning the Protein Language: Evolution, Structure and Function > Supervision encodes biological meaning
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
- Rests on an assumption: Proteins with similar structures and functions still exist below 30% sequence identity, but they can no longer be detected from sequence similarity alone and it is unclear whether their functions are conserved.
- Rests on an assumption: Evolutionary innovation requires that protein function can change with only a few mutations.
None of the sources cited for this claim have been checked yet.
In the paper
“Language models learn the probability of a sequence occurring and this can be directly applied to predict the fitness of sequence mutations 29-31 .”
§ Learning the Protein Language: Evolution, Structure and Function > Introduction
What the sources say
- Not checked yetcites [29] Riesselman AJ, Ingraham JB & Marks DS Deep generative models of genetic variation capture the effects of mutations. Nat. Methods 15, 816– 822 (2018). 30250057 PMC6693876 source
- Not checked yetcites [30] Hie B, Zhong E, Bryson B & Berger B Learning mutational semantics. Adv. Neural Inf. Process. Syst 33, (2020).
- Not checked yetcites [31] Hie B, Zhong ED, Berger B & Bryson B Learning the language of viral evolution and escape. Science 371, 284– 288 (2021). 33446556 source
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt 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
“Language models capture complex relationships between residues in protein sequences by condensing information from enormous protein sequence databases.”
§ Learning the Protein Language: Evolution, Structure and Function > Conclusions and Perspectives: Strong biological priors are key to improving protein language models
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✗ Based: failsProbably not a substantial claim. The paper may give no reason why it holds.
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
“Fine-tuning pre-trained models can help with this problem but does not fully resolve it.”
§ Learning the Protein Language: Evolution, Structure and Function > Conclusions and Perspectives: Strong biological priors are key to improving protein language models
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesProbably not a substantial claim. It may state two separate claims as one.
- Rests on an assumption: For individual protein families or application-specific protein datasets, brute force hyperparameter search is the gold standard for selecting model architectures and number of parameters.
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
“High compute costs and murky design guidelines mean that developing new models is often an expensive, time consuming, and ad hoc process.”
§ Learning the Protein Language: Evolution, Structure and Function > Conclusions and Perspectives: Strong biological priors are key to improving protein language models
Strength:hedged· not yet calibratedCausal:direct· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✗ Load-bearing: fails✓ Based: passesProbably not a substantial claim. The paper’s conclusions may not depend on it.
None of the sources cited for this claim could be matched to a paper in Crossref, so there is nothing to check it against.
In the paper
“To compare methods, we follow Luo et al and perform 5-fold cross validation on each deep mutational scanning dataset 34”
§ Learning the Protein Language: Evolution, Structure and Function > Methods > Sequence-to-phenotype prediction and evaluation
What the sources say
- Not matchedcites Luo et al
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- Not matchedcites [34] Luo Y Evolutionary Context-Integrated Deep Sequence Modeling for Protein Engineering. in Research in Computational Molecular Biology 261– 263 (Springer International Publishing, 2020).
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
Load-bearingStrength:plain· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesProbably not a substantial claim.
- Not matched
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 compare methods by 10-fold cross validation.”
§ Learning the Protein Language: Evolution, Structure and Function > Methods > Transmembrane region prediction training and evaluation
Load-bearingStrength:plain· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesProbably not a substantial claim.
None of the sources cited for this claim have been checked yet.
In the paper
“New linear complexity attention mechanisms may help to alleviate this limitation 95,96.”
§ Learning the Protein Language: Evolution, Structure and Function > Conclusions and Perspectives: Strong biological priors are key to improving protein language models
What the sources say
- Not matchedcites [95] Choromanski KM Rethinking Attention with Performers. in International Conference on Learning Representations (2020).
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- Not checked yetcites [96] Wang S, Li BZ, Khabsa M, Fang H & Ma H Linformer: Self-Attention with Linear Complexity. (2020). source
Strength:hedged· not yet calibratedCausal:conditional· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✗ Load-bearing: fails✓ Based: passesProbably not a substantial claim. The paper’s conclusions may not depend on it.
- Rests on an assumption: Transformers scale quadratically with sequence length.
- Not matched
In the paper
“Recently, Luo et al. presented a method for combining language model-based representations with local evolutionary context-based representations (ECNet) and demonstrated that these representations were powerful for sequence-to-phenotype mapping on a panel of deep mutational scanning datasets 34.”
§ Learning the Protein Language: Evolution, Structure and Function > Transfer learning improves downstream applications
What the sources say
- Not matchedcites [34] Luo Y Evolutionary Context-Integrated Deep Sequence Modeling for Protein Engineering. in Research in Computational Molecular Biology 261– 263 (Springer International Publishing, 2020).
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
Strength:boosted· not yet calibratedCausal:correlational· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✗ Load-bearing: fails✓ Based: passesProbably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.
- Not matched
The paper relies on this without support.
Many problems in biology are small data problems; quantitative protein characterization assays are rarely high throughput and methods are needed that can generalize given only 10s to 100s of experimental measurements.
“A key challenge in biology is that many problems are small data problems. Quantitative protein characterization assays are rarely high throughput and methods are needed that can generalize given only 10s to 100s of experimental measurements.”
The paper relies on this without support.
We are often interested in extrapolating from data collected over a small region of protein sequence space to other sequences, often with little to no homology.
“Furthermore, we are often interested in extrapolating from data collected over a small region of protein sequence space to other sequences, often with little to no homology.”
The paper relies on this without support.
Proteins with similar structures and functions still exist below 30% sequence identity, but they can no longer be detected from sequence similarity alone and it is unclear whether their functions are conserved.
“Often called the “twilight zone” of protein sequence homology, proteins with similar structures and functions still exist below this level, but they can no longer be detected from sequence similarity alone and it is unclear whether their functions are conserved.”
The paper relies on this without support.
Evolutionary innovation requires that protein function can change with only a few mutations.
“Evolutionary innovation requires that protein function can change with only a few mutations.”
The paper relies on this without support.
Proteins exist in physical space.
“Proteins are objects that exist in physical space.”
The paper relies on this without support.
Many fundamental evolutionary processes that give rise to the diversity of protein sequences observed today are understood.
“Similarly, we understand many of the fundamental evolutionary processes that give rise to the diversity of protein sequences observed today.”
The paper relies on this without support.
Physics and evolution are key properties of proteins.
“These two elements, physics and evolution, are the key properties of proteins and our models might benefit from being structured explicitly to incorporate evolutionary and physics-based inductive biases.”
The paper relies on this without support.
Biological sequences are not natural language.
“Biological sequences are not natural language”
The paper relies on this without support.
Filtering sequences based on percent identity is a conventional crude address to the phylogenetic assumption problem.
“Conventionally, this problem is crudely addressed by filtering sequences based on percent identity.”
The paper relies on this without support.
At the upper limit of large natural protein databases, larger models continue to yield improved performance.
“At the upper limit of large natural protein databases, larger models continue to yield improved performance.”
The paper relies on this without support.
For individual protein families or application-specific protein datasets, brute force hyperparameter search is the gold standard for selecting model architectures and number of parameters.
“However, for individual protein families or other application specific protein data sets, the gold standard is to select model architectures and number of parameters via brute force hyperparameter search methods.”
The paper relies on this without support.
Transformers scale quadratically with sequence length.
“Transformers scale quadratically with sequence length”
The paper relies on this without support.
Recurrent neural networks scale linearly with sequence length.
“This problem is less extreme for recurrent neural networks, which scale linearly with sequence length”
The paper relies on this without support.
Increasing model size, compute power, and dataset size will continue to improve performance of protein language models.
“Increasing model size, compute power, and dataset size will only continue to improve performance of protein language models.”
The paper relies on this without support.
Protein language models are transforming computational protein biology today due to their ease of use and widespread applicability.
“Already, these methods are transforming computational protein biology today due to their ease of use and widespread applicability.”
The paper relies on this without support.
Augmenting language models with protein-specific properties such as structure and function offers an already successful route towards richer representations and novel biology.
“Furthermore, augmenting language models with protein specific properties such as structure and function offers one already successful route towards even richer representations and novel biology.”
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?
In the citing paper
“Instead of using multi-headed self-attention (MHSA), we adopt the Gated Attention Unit (GAU) (line 8 in Algorithm 1), which has shown great promise as an alternative to MHSAs with smaller memory consumption and faster convergence rate (25).”
From High-resolution de novo structure prediction from primary sequence (2022) · cites this paper as [25]
In the citing paper
“The original GAU in (25) also suggests that relu 2 (•) performs better compared to softmax in terms of both computation speed and convergence rate.”
From High-resolution de novo structure prediction from primary sequence (2022) · cites this paper as [25]
Citations
Cites
Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, L., & Polosukhin, I. (2017) · arXiv · doi:10.48550/arxiv.1706.03762 · Cited by 120,404
Hochreiter, S., & Schmidhuber, J. (1997) · Neural Computation · doi:10.1162/neco.1997.9.8.1735 · Cited by 102,318
Adam: A Method for Stochastic Optimization
Kingma, D. P., & Ba, J. (2014) · arXiv · doi:10.48550/arxiv.1412.6980 · Cited by 98,946
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
Devlin, J., Chang, M.-W., Lee, K., & Toutanova, K. (2018) · arXiv · doi:10.48550/arxiv.1810.04805 · Cited by 73,173
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
Devlin, J., Chang, M.-W., Lee, K., & Toutanova, K. (2019) · Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers) · doi:10.18653/v1/n19-1423 · Cited by 54,325
Berman, H. M. (2000) · Nucleic Acids Research · doi:10.1093/nar/28.1.235 · Cited by 40,196
ImageNet Large Scale Visual Recognition Challenge
Russakovsky, O., Deng, J., Su, H., Krause, J., Satheesh, S., Ma, S., Huang, Z., Karpathy, A., Khosla, A., Bernstein, M., et al. (2015) · International Journal of Computer Vision · doi:10.1007/s11263-015-0816-y · Cited by 39,380
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
MRBAYES: Bayesian inference of phylogenetic trees.
Huelsenbeck, J. P., & Ronquist, F. (2001) · Bioinformatics · doi:10.1093/bioinformatics/17.8.754 · Cited by 21,937
GROMACS 4: Algorithms for Highly Efficient, Load-Balanced, and Scalable Molecular Simulation
Hess, B., Kutzner, C., van der Spoel, D., & Lindahl, E. R. (2008) · Journal of Chemical Theory and Computation · doi:10.1021/ct700301q · Cited by 15,830
Flagged first, then by citations
3 of the 133 references couldn’t be matched to a known paper.
Research works printed but not linked: 4
- Radford, 2019, Language models are unsupervised multitask learners, OpenAI blog, 1
- Remmert, 2011, HHblits: lightning-fast iterative protein sequence searching by HMM-HMM alignment, Nat. Methods, 9, 173, 10.1038/nmeth.1818 · DOI · Paper page
- Remmert, 2011, HHblits: lightning-fast iterative protein sequence searching by HMM-HMM alignment, Nat. Methods, 9, 173, 10.1038/nmeth.1818 · DOI · Paper page
- Tsirigos, 2015, The TOPCONS web server for consensus prediction of membrane protein topology and signal peptides, Nucleic Acids Res., 43, W401-7, 10.1093/nar/gkv485 · DOI
Cited by
Sorted by citations
Items 301–400 of 486
Deep learning on protein language model embeddings unlocks accurate prediction of protein solubility
Cui, J., Jiang, Y., Wang, J., Jiang, H., Fang, J., Jiang, J., Chen, Q., Zhong, S., & Wang, X. (2026) · Frontiers in Microbiology · doi:10.3389/fmicb.2026.1716930 · Cited by 3
“…to capture complex long-range dependencies, evolutionary signals, and structural-functional relationships embedded within sequences, enabling the capabilities that far surpass traditional encodings like one-hot or handcrafted physicochemical descriptors (Bepler and Berger, 2021; Lupo et al., 2022).”
A tri-modal contrastive learning framework for protein representation learning
Zhang, L., Guo, H., Schaffer, L. V., Ko, Y. S., Singh, D., Rahmani, H., Grotjahn, D. A., Villa, E., Gilson, M., Wang, W., et al. (2026) · Cell Reports Methods · doi:10.1016/j.crmeth.2026.101407 · Cited by 6
“3 15 These models, particularly protein language models (PLMs), 5–8, 10, 13 have prediction, 18,19 structure prediction, 20,21 and protein design.”
Advancements in characterization of protein dynamics with machine learning
Madani, M., & Tarakanova, A. (2026) · npj Soft Matter · doi:10.1038/s44431-026-00025-w · Cited by 3
“…strategies — autoregressive, bidirectional, and masked modeling — are commonly employed to extract sequence-based patterns that inform structure prediction 199 , and their reduced dependence on MSAs makes them useful for families with sparse homology or pronounced conformational variability.”
Zheng, W. (2024) · PLoS ONE · doi:10.1371/journal.pone.0302504 · Cited by 3
Generative modeling for RNA splicing prediction and design
Wu, D., Maus, N., Jha, A., Yang, K., Wales-McGrath, B. D., Jewell, S., Tangiyan, A., Choi, P. S., Gardner, J. R., & Barash, Y. (2026) · eLife · doi:10.7554/elife.106043.3 · Cited by 3
Dango: Predicting higher-order genetic interactions
Zhang, R., Bafna, M., Ma, J., & Ma, J. (2026) · Cell Systems · doi:10.1016/j.cels.2026.101593 · Cited by 3
PESI: Paratope-Epitope Set Interaction for SARS-CoV-2 Neutralization Prediction
Wan, Z., Lin, Z., Rashid, S., Ng, S. Y.-H., Yin, R., Senthilnath, J., & Kwoh, C. K. (2023) · IEEE International Conference on Bioinformatics and Biomedicine (BIBM) · doi:10.1109/bibm58861.2023.10386059 · Cited by 3
“Besides, scaling language models with massive amino acid sequences also derives structural knowledge based on merely sequence inputs [3], [11], [39], which also demonstrates its capacity in modeling protein sequences.”
BackgroundKalantar, M., Kalanther, I., Kumar, S., Buxton, E. K., & Raeeszadeh‐Sarmazdeh, M. (2024) · Computational and Structural Biotechnology Journal · doi:10.1016/j.csbj.2024.10.005 · Cited by 3
“These developments have led to the creation of Large Protein Language Models (LPLMs), which assist in discovering the evolutionary, structural, and functional properties across protein space by encoding amino-acid sequences into numeric vector representations [36].”
Zhang, Y., Deng, J., Dong, M., Wu, J., Zhao, Q., Gao, X., & Xiong, D. (2025) · Neural Networks · doi:10.1016/j.neunet.2025.107476 · Cited by 3
CELL-E: A Text-To-Image Transformer for Protein Localization PredictionPreprint
Khwaja, E., Song, Y. S., & Huang, B. (2022) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2022.05.27.493774 · Cited by 4
“We also experimented with other types of protein embeddings, including UniRep [30] and ESM1-b [31].”
BackgroundMethodDeep Local Analysis estimates effects of mutations on protein-protein interactionsPreprint
Behbahani, Y. M., Laine, É., & Carbone, A. (2022) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2022.10.09.511484 · Cited by 3
“Under review. and functional properties in the context of protein language models [30, 12, 3], for fixed-backbone protein design [1, 17, 8], and for protein stability predictions [5, 39].”
BackgroundContrasting Sequence with Structure: Pre-training Graph Representations with PLMsPreprint
Robinson, L., Atkinson, T. J., Copoiu, L., Bordes, P., Pierrot, T., & Barrett, T. D. (2023) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2023.12.01.569611 · Cited by 3
“…have demonstrated the ability to capture intrinsic relationships in amino acid sequences [1] in rich representations that can be leveraged for various downstream applications [2] - mirroring the trends observed in other domains such as natural language processing [3, 4] and computer vision [5].”
BackgroundKellman, B. P., Mariethoz, J. A., Zhang, Y., Shaul, S., Alteri, M., Sandoval, D. R., Jeffris, M., Armingol, E., Bao, B., Lisacek, F., et al. (2024) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2024.05.15.594334 · Cited by 3
Learning a CoNCISE language for small-molecule bindingPreprint
Erden, M., Devkota, K., Varghese, L., Cowen, L., & Singh, R. (2025) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2025.01.08.632039 · Cited by 5
Generative modeling for RNA splicing predictions and designPreprint
Wu, D., Maus, N., Jha, A., Yang, K., Wales-McGrath, B. D., Jewell, S., Tangiyan, A., Choi, P. S., Gardner, J. R., & Barash, Y. (2025) · eLife · doi:10.7554/elife.106043.1 · Cited by 4
PLANET-MD: Ultra-fast Proteome-scale Prediction of Allosteric Networks in ProteinsPreprint
Sledzieski, S., & Hanson, S. M. (2025) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2025.06.12.659353 · Cited by 6
From Signal to Symphony: Exploring 2D Sequence Representations for Protein Function Prediction
Wang, Y., Cai, M., Dong, Y., Ma, Y., & Wei, K. (2025) · Journal of Chemical Information and Modeling · doi:10.1021/acs.jcim.5c01768 · Cited by 3
NeuroPpred-MSN: A Neuropeptide Prediction Model Based on Multi-feature Fusion and Siamese Networks
Wen, J., Chen, M., Shen, Y., Wang, H., Wei, Z., Gu, L., & Zhu, X. (2025) · Interdisciplinary Sciences Computational Life Sciences · doi:10.1007/s12539-025-00730-6 · Cited by 2
“In recent years, pre-trained protein language models have been used for bioactive peptide prediction [3, 21–24].”
Si, W., Chen, H., & Wu, G. (2025) · Physical Chemistry Chemical Physics · doi:10.1039/d5cp01017a · Cited by 2
Su, Y., Guo, Q., & Liu, T. (2025) · Interdisciplinary Sciences Computational Life Sciences · doi:10.1007/s12539-025-00766-8 · Cited by 2
“These pre-trained PLMs include ESM-2 [34], ESM-1b [50], ProtT5 [35], ProSE [51], and Tape [52].”
TCR representation learning with protein language models: a comprehensive review
Kinoshita, K., & Kobayashi, T. J. (2025) · International Immunology · doi:10.1093/intimm/dxaf048 · Cited by 3
“These models effectively extract structural and functional protein features through self-supervised learning on large amino acid sequence datasets (41).”
Khatri, K., Levy, R. M., & Haldane, A. (2025) · Physical Review Research · doi:10.1103/g5cx-1vhj · Cited by 3
“To function well, GPSMs must capture covarying amino acid patterns in the MSA that implicitly encode information about physical constraints on proteins, Published by the American Physical Society under the terms of the Creative Commons Attribution 4.0 International license.”
InfluentialPLM-eXplain: divide and conquer the protein embedding space
van Eck, J., Gogishvili, D., Silva, W., & Abeln, S. (2025) · Bioinformatics · doi:10.1093/bioinformatics/btaf631 · Cited by 2
“These models quickly outperformed existing tools across a variety of protein prediction tasks (Bepler and Berger 2021, Elnaggar et al. 2022, Lin et al. 2023, Yu et al. 2023, Hie et al. 2024, Hayes et al. 2025), enabling highly accurate predictions of properties ranging from secondary structure…”
Learnable protein representations in computational biology for predicting drug-target affinity
Kumar, R., Romano, J. D., & Ritchie, M. D. (2026) · Journal of Cheminformatics · doi:10.1186/s13321-025-01145-7 · Cited by 2
“…broad field of protein representation learning, where researchers try to generate some meaningful representation of a protein (often in some lower-dimensional, continuous latent space) that can be used for a specific downstream task, for a generalized set of tasks, or a combination of both [3–5].”
Wu, F., Wu, R., Chen, L., Chen, Q., & Liu, H. (2026) · New Biotechnology · doi:10.1016/j.nbt.2026.01.009 · Cited by 3
Ghosh, S., & Mitra, P. (2026) · IEEE Access · doi:10.1109/access.2026.3658221 · Cited by 2
Protein language models enable accurate viral host range prediction
Beltrán, J. F., Belén, L. H., Contreras, F. P., & Yáñez, A. J. (2026) · Scientific Reports · doi:10.1038/s41598-026-37765-8 · Cited by 2
Li, S., Wang, Y., Zhang, H., Xing, X., Shao, Y., & Qi, Z. (2026) · BMC Genomics · doi:10.1186/s12864-026-12762-3 · Cited by 2
Özbulak, U., Lee, H. J., Zuallaert, J., Neve, W. D., Depuydt, S., & Vankerschaver, J. (2023) · Bioinformatics · doi:10.1093/bioinformatics/btad338 · Cited by 2
“…such as acceptor, donor, and termination sites which are all integral to the creation of a functional protein but not to translation initiation exclusively can divert the model’s attention from translation initiation to these additional factors (Jaganathan et al. 2019, Bepler and Berger 2021).”
PreHom-PCLM: protein remote homology detection by combing motifs and protein cubic language model
Shao, J., Zhang, Q., Yan, K., & Liu, B. (2023) · Briefings in Bioinformatics · doi:10.1093/bib/bbad347 · Cited by 2
Ji, Y., Shang, J., Guan, J., Zou, W. H., Liao, H., Tang, X., & Sun, Y. (2024) · GigaScience · doi:10.1093/gigascience/giae104 · Cited by 2
Explaining Time Series Classification Predictions via Causal Attributions
Alcaraz, J. M. L., & Strodthoff, N. (2025) · IEEE International Conference on Tools with Artificial Intelligence (ICTAI) · doi:10.1109/ictai66417.2025.00014 · Cited by 3
“Time series data, prevalent in natural sciences, medicine, and life sciences [1]–[5], enable modeling of temporal patterns, especially for classification tasks [6], [7].”
Evo-velocity: Protein language modeling accelerates the study of evolution
Sandhu, M., Spence, M. A., & Jackson, C. J. (2022) · Cell Systems · doi:10.1016/j.cels.2022.03.004 · Cited by 2
A Comparative Analysis of Transformer-based Protein Language Models for Remote Homology Prediction
Kabir, A., Moldwin, A., & Shehu, A. (2023) · ACM International Conference on Bioinformatics, Computational Biology, and Health Informatics (ACM-BCB) · doi:10.1145/3584371.3612942 · Cited by 2
“An explosion in the number of known protein sequences is allowing researchers to leverage the Transformer [29] architecture and build Protein Language Models (PLMs) [4, 11, 13].”
BackgroundZhang, G., Wang, H., Zhang, Z., Zhang, L., Guo, G., Yang, J., Yuan, F., & Ju, F. (2022) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2022.12.28.522150 · Cited by 2
Evolutionary selection of proteins with two folds
Schafer, J. W., & Porter, L. L. (2023) · doi:10.1101/2023.01.18.524637 · Cited by 2
Do Domain-Specific Protein Language Models Outperform General Models on Immunology-Related Tasks?Preprint
Deutschmann, N., Pélissier, A., Weber, A., Gao, S., Bogojeska, J., & Martínez, M. R. (2023) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2023.10.17.562795 · Cited by 4
Mutation Pathogenicity Prediction by a Biology Based Explainable AI Multi-Modal AlgorithmPreprint
Kellerman, R., Nayshool, O., Barel, O., Paz, S., Amariglio, N., Klang, E., & Rechavi, G. (2024) · medRxiv · doi:10.1101/2024.06.05.24308476 · Cited by 2
“Ensemble learning 26 principal components N as principal components analysis (PCA), retaining the top offers another straight-forward tactic, with the adjustment of the weights of the predictions retrieved from the discrete models by e.g. logistic regression.”
MethodDeep mutational scanning reveals sequence to function constraints for SWEET family transportersPreprint
Narayanan, K. K., Weigle, A. T., Xu, L., Mi, X., Zhang, C., Chen, L., Procko, E., & Shukla, D. (2024) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2024.06.28.601307 · Cited by 3
“Recently developed pretrained language model embeddings can be calculated without deriving sequence profiles, 38,39 creating opportunity for researchers to predict variant effects within and across plant proteomes.”
Ji, Y.-X., Shang, J., Guan, J.-J., Zou, W., Liao, H., Tang, X.-B., & Sun, Y.-N. (2024) · bioRxiv · doi:10.1101/2024.07.03.602011 · Cited by 4
“In addition, we will leverage protein language models ( PLM s), which have demonstrated remarkable performance in various protein-related tasks by understanding the underlying semantic meaning of the language of life [16].”
MethodGenerative modeling for RNA splicing prediction and designPreprint
Wu, D., Maus, N., Jha, A., Yang, K., Wales-McGrath, B. D., Jewell, S., Tangiyan, A., Choi, P. S., Gardner, J. R., & Barash, Y. (2025) · eLife · doi:10.7554/elife.106043 · Cited by 3
Deep Learning-coupled Proximity Proteomics to Deconvolve Kinase Signaling In VivoPreprint
Jha, K., Shonai, D., Parekh, A., Uezu, A., Fujiyama, T., Yamamoto, H., Parameswaran, P., Yanagisawa, M., Singh, R., & Soderling, S. H. (2025) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2025.04.27.650849 · Cited by 3
“More recently, large protein language models (PLMs) have revolutionized the field of protein biology by capturing the intricate dependencies within sequences using transformer-based architectures ( 27-29 ).”
GOLF: A Generative AI Framework for Pathogenicity Prediction of Myocilin OLF VariantsPreprint
Walton, T., Tsui, D., Fogel, L., Huard, D. J. E., Chagas, R. S., Lieberman, R. L., & Aghazadeh, A. (2025) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2025.06.17.660210 · Cited by 2
“Recent studies have shown that generative models trained on evolutionary sequences implicitly capture structural and biochemical constraints from sequence variations alone 16,33,34 .”
Abbass, J., & Parisi, C. (2024) · Journal of Biomolecular Structure and Dynamics · doi:10.1080/07391102.2024.2328736 · Cited by 1
“…both classification hierarchies have been gaining much attention; this three-actor relationship plays a critical role in many fields in protein biology such as understanding disease mechanisms and designing proteins and drugs for therapeutic and bioengineering applications (Bepler & Berger, 2021).”
InfluentialDong, K., & Gao, J. (2024) · Current Bioinformatics · doi:10.2174/0115748936291075240409080924 · Cited by 1
Progress in protein pre-training models integrating structural knowledge
Tang, T.-Y., Xiong, Y., Zhang, R.-G., Zhang, J., Li, W., Wang, J., & Wang, W. (2024) · Acta Physica Sinica · doi:10.7498/aps.73.20240811 · Cited by 2
Progress and challenges for the application of machine learning for neglected tropical diseases
Khew, C., Akbar, R., & Mohd‐Assaad, N. (2025) · F1000Research · doi:10.12688/f1000research.129064.2 · Cited by 2
Peng, S., Lyu, L., Registre, L., Subedi, S., Mubassir, M. H. M., Bakheet, M., & Bahl, J. (2025) · Frontiers in Immunology · doi:10.3389/fimmu.2025.1576414 · Cited by 1
“Protein language models (PLMs) offer a novel perspective for studying viral evolution by treating amino acid sequences as analogous to sentences, leveraging large-scale sequence data to capture the biochemical properties of proteins (11).”
Machine Learning for Protein Function Prediction
Zhu, Y.-H., Liu, Z., Ding, Y. Q., Ji, Z., & Yu, D.-J. (2025) · Methods in molecular biology · doi:10.1007/978-1-0716-4662-5_2 · Cited by 1
Predicting Subcutaneous Antibody Bioavailability Using Ensemble Protein Language Models.
Cabreza, M., Hojegian, W., Wu, I.-E., & Lai, P.-K. (2025) · Molecular Pharmaceutics · doi:10.1021/acs.molpharmaceut.5c00523 · Cited by 2
Progress and challenges for the application of machine learning for neglected tropical diseases
Khew, C., Akbar, R., & Mohd‐Assaad, N. (2025) · F1000Research · doi:10.12688/f1000research.129064.3 · Cited by 1
Zhang, R., Wu, H., Liu, C., Li, H., Wu, Y., Li, K., Wang, Y., Deng, Y., Chen, J., Zhou, F., et al. (2025) · Neural Networks · doi:10.1016/j.neunet.2025.108148 · Cited by 1
Predicting Peptide Aggregation with Protein Language Model Embeddings
Eschbach, E., Deibler, K., Korani, D., & Swanson, S. (2026) · Journal of Chemical Information and Modeling · doi:10.1021/acs.jcim.6c00577 · Cited by 1
CATHe2: Enhanced CATH superfamily detection using ProstT5 and structural alphabets
(2025) · doi:10.1093/biomethods/bpaf080 · Cited by 1
“All this structural information can be inferred information or ground truth and used as is or embedded using pLM for more complex feature extraction [35–38 , 25].”
Ma, X., Dey, S., Vaishnavey, S. R., Zelinski, C., Li, Q., & Chowdhury, R. (2025) · NAR Genomics and Bioinformatics · doi:10.1093/nargab/lqaf154 · Cited by 3
“This limitation is partly addressed by contemporary PLMs that integrate structural and functional information in their training [ 64 ].”
Song, J., Geng, A., Wen, Q., Xu, J., Meng, Y., Cui, F., Zou, Q., Wei, L., & Zhang, Z. (2025) · Journal of Chemical Information and Modeling · doi:10.1021/acs.jcim.5c02196 · Cited by 2
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
Chen, J., Zhang, G., Xu, Z., & Zhang, Q. (2026) · Briefings in Bioinformatics · doi:10.1093/bib/bbag067 · Cited by 1
“They have increasingly benefited from protein language models (PLMs) [14, 15], which provide context-aware embeddings that improve over handcrafted s equence descriptors.”
Identifying microbial protease allergens through protein language model-guided homology
Thurimella, K., Wu, E., Li, C., Graham, D. B., Owens, R. M., Plichta, D. R., Sokol, C. L., Xavier, R. J., & Bacallado, S. (2026) · Cell Systems · doi:10.1016/j.cels.2025.101510 · Cited by 1
Li, J., Kang, X., Su, C., Shi, H., Cui, F., Zhang, Z., Lin, C., & Wei, L. (2026) · ACS Synthetic Biology · doi:10.1021/acssynbio.6c00015 · Cited by 2
Zhou, Q., Sun, L., Ou, Y., Liu, H., Gao, J., Dai, R., Li, Y., Cheng, G., Tao, Y., Pan, Y., et al. (2026) · Briefings in Bioinformatics · doi:10.1093/bib/bbag107 · Cited by 1
“Machine learning (ML) and deep learning (DL) algorithms, such as protein structure prediction [12–14], virtual screening [15–19], toxicity prediction [20–22], pharmacophore modeling [23–26], drug r epositioning [27–34], and clinical monitoring [35–46], have been applied to drug discovery.”
Ye, Y., Xie, P., Chung, C., Wang, F., Shi, B., Lin, X., Chiu, Y.-P., Lee, T.-Y., Liu, X., Guan, J., et al. (2026) · International Journal of Biological Macromolecules · doi:10.1016/j.ijbiomac.2026.152152 · Cited by 1
He, H., Yang, F.-S., Zhang, Y.-X., Xu, C., Ma, X. Q., Song, Y., Sun, J., & Yu, Y. (2026) · Annals of Botany · doi:10.1093/aob/mcag114 · Cited by 1
Protein language visualizer: a repository for homology exploration with language model embeddings
Espinoza-Herrera, J., Manríquez-García, M. F., Bermejo, S. M., Jasso, A. L., Ruiz-Alcocer, J. P., Siordia, A., Veskimägi, S. M., Roethler, N., & Jinich, A. (2026) · Digital Discovery · doi:10.1039/d5dd00472a · Cited by 2
Generative artificial intelligence for enzyme design and biocatalysis
Middendorf, L., & Ferruz, N. (2026) · Current Opinion in Chemical Biology · doi:10.1016/j.cbpa.2026.102728 · Cited by 3
Sequence-Structure Embeddings via Protein Language Models Improve on Prediction Tasks
Kabir, A., & Shehu, A. (2022) · IEEE International Conference on Knowledge Graph (ICKG) · doi:10.1109/ickg55886.2022.00021 · Cited by 1
“Work in [3] proposes a PLM based on bidirectional LSTM with structural supervision.”
BackgroundQin, X., Liu, M., & Liu, G. (2023) · Computers in Biology and Medicine · doi:10.1016/j.compbiomed.2023.107571 · Cited by 2
Sui, J., Chen, J., Chen, Y., Iwamori, N., & Sun, J. (2023) · Frontiers in Bioscience-Landmark · doi:10.31083/j.fbl2812322 · Cited by 1
“Training protein language models by self-supervised learning of large amounts of natural sequence data and structural supervision of smaller sequence sets [22].”
BackgroundMethodLiu, T., Song, C., & Wang, C. (2023) · Mathematical Biosciences & Engineering · doi:10.3934/mbe.2024063 · Cited by 1
“(3) The structure of ProSE is a three-layer BiLSTM similar to the SeqVec structure, with the difference that it uses not only the sequence data but also the structural information of the proteins [19].”
BackgroundMethodCELL-E: A Text-to-Image Transformer for Protein Image Prediction
Khwaja, E., Song, Y. S., & Huang, B. (2024) · Lecture notes in computer science · doi:10.1007/978-1-0716-3989-4_12 · Cited by 1
How Much Do DNA and Protein Deep Embeddings Preserve Biological Information?
Tolloso, M., Galfrè, S. G., Pavone, A., Podda, M., Sîrbu, A., & Priami, C. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-71671-3_15 · Cited by 1
Peptide Sequencing Via Protein Language Models
Pham, T. L. H., Saurav, J. R., Omere, A. A., Heyl, C. J., Nasr, M. S., Reynolds, C. R., Veerla, J. P., Shang, H. H., Jaworski, J. W., Ravenscraft, A., et al. (2024) · ACM International Conference on Bioinformatics, Computational Biology, and Health Informatics (ACM-BCB) · doi:10.1145/3698587.3701385 · Cited by 3
“Large language models (LLMs) have shown remarkable adaptability in interpreting protein sequences, excelling in predicting structures, functions, and evolutionary relationships [7, 35, 42].”
SCOP: A Sequence-Structure Contrast-Aware Framework for Protein Function Prediction
Ma, R., He, C., Zheng, H., Wang, X., Wang, H., Zhang, Y., & Duan, L. (2024) · IEEE International Conference on Bioinformatics and Biomedicine (BIBM) · doi:10.1109/bibm62325.2024.10822541 · Cited by 1
“Most methods [4], [8], [9] leverage language models to learn representations from protein sequences and then predict their functions, whereas others [10], [11] learn the structural features of proteins to predict their functions.”
MuFAl: A Universal Drug-Target Interaction Prediction Framework
Chen, X., Husen, M. N., & Huang, X. (2025) · International Conference on Ubiquitous Information Management and Communication (IMCOM) · doi:10.1109/imcom64595.2025.10857588 · Cited by 1
“The former contains amino acid sequence information and partial structural information of proteins [9], performing in experiments better than Prose [33] and ESM [34].”
Deep Learning-Enhanced Gap Filling in Drosophila Melanogaster Genomic Data
Sharma, J., Jetschny, S., Kapun, M., & Belaid, M. B. (2024) · International Conference on Machine Learning and Applications · doi:10.1109/icmla61862.2024.00224 · Cited by 1
“CpG Transformer adapts the Transformer architecture for DNA methylation imputation by leveraging attention mechanisms to learn from large CpG site matrices [1].”
Draizen, E. J., Mura, C., & Bourne, P. E. (2022) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2022.11.16.516787 · Cited by 1
“While many new sequence-based deep learning methods, based on large language models [2, 3, 4, 5], can identify more remote similarities than can hidden Markov models (HMMs) or classic sequence-comparison algorithms (e.g., BLAST), larger structural rearrangements and permutations, such as occur on…”
BackgroundDomain loss enabled evolution of novel functions in a gene superfamilyPreprint
Koludarov, I., Senoner, T., Jackson, T., Dashevsky, D., Heinzinger, M., Aird, S. D., & Rost, B. (2022) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2022.12.15.520616 · Cited by 1
ZIPPI: proteome-scale sequence-based evaluation of protein-protein interaction modelsPreprint
Zhao, H., Petrey, D. S., Murray, D., & Honig, B. H. (2023) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2023.05.25.542344 · Cited by 1
Opinion piece/Book chapter: Artificial Intelligence learns protein predictionPreprint
Heinzinger, M., & Rost, B. (2023) · OSF Preprints (OSF Preprints) · doi:10.31219/osf.io/n7wfp · Cited by 1
Homology detection using a protein secondary structure-based large language modelPreprint
Kogay, R., Ma, W., Bousselham, J., Yang, Z., Rockmore, D. N., Zhaxybayeva, O., & Vosoughi, S. (2023) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2023.12.19.572443 · Cited by 1
Prediction of protein-carbohydrate binding sites from protein primary sequencePreprint
Nafi, M. M. I., Nawar, Q. F., Islam, T. N., & Rahman, M. S. (2024) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2024.02.09.579590 · Cited by 1
“An increasingly successful trend in various protein attribute prediction problems has been the utilization of recently developed protein language models [21, 22].”
Eberhardt, J., Lees, A., Lill, M. A., & Schwede, T. F. (2024) · ChemRxiv · doi:10.26434/chemrxiv-2023-b7l81-v2 · Cited by 1
Protein sequence domain annotation using a language modelPreprint
Sarkar, A., Krishnan, K., & Eddy, S. R. (2024) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2024.06.04.596712 · Cited by 2
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
“Protein Language Models (pLMs), also adapted from NLP, are the conceptual successors of HMMs [21, 22].”
A Deep Learning Method for Predicting Interactions for Intrinsically Disordered Regions of ProteinsPreprint
Majila, K., Ullanat, V., & Viswanath, S. (2025) · SSRN Electronic Journal · doi:10.2139/ssrn.5133412 · Cited by 1
Rayakar, A. A., Jaladanki, C. K., Ong, Q., Tan, B., Chen, J., Tan, A. Q. L., Lim, L. T. R., Kohaal, N., Chen, Y., Wang, J., et al. (2025) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2025.03.16.643501 · Cited by 1
“Two of the constituent feature sets are embeddings derived from protein language models (pLMs) which convert protein sequences into per-amino acid distributed vector representations 20 .”
Evaluating the Effectiveness of Parameter-Efficient Fine-Tuning in Genomic Classification TasksPreprint
Berman, D., Jimenez, D., Ta, S., Merritt, B. B., Ratcliff, J., & Narayan, V. (2025) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2025.08.21.671544 · Cited by 1
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
“…models (pLMs) such as ProteinBERT (3), the ProtTrans model family (4), and the Evolutionary Scale Modeling (ESM) series (5–7) have emerged as a transformative paradigm for learning context-aware representations of amino acid sequences with a variety of biological and clinical applications (8–10).”
Srivastava, H. R., Berenberg, D., Qassab, O., Wang, Z., Bonneau, R., & Carlton, J. M. (2025) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2025.09.12.675843 · Cited by 1
Fast, accurate construction of multiple sequence alignments from protein language embeddingsPreprint
Hoang, M., Armour-Garb, I., & Singh, M. (2026) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.64898/2026.01.02.697423 · Cited by 1
“For each model, ARIES SP scores on BAliBASE are computed when representing amino acids using concatenations of the last ℓ PLM layer embeddings, with ℓ ∈ [1 , 3 , 6 , 9 , 12].”
Contrastive learning for antibody-antigen sequence-to-specificity predictionPreprint
Lee, H., Castro, K. M., Renwick, S., Stalder, L., Glänzer, W. S., Kumar, R. K., Chen, N., Scheck, A., Yermanos, A., Mason, D. M., et al. (2026) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.64898/2026.02.25.707916 · Cited by 3
“8 .”
(2026) · doi:10.1073/pnas.2524201123 · Cited by 2
“This allows our model to leverage the rich functional sequence-derived properties (63, 64) learned for each protein by protein language models (3, 9, 10, 14) (see also refs.”
Pre-training Protein Models with Molecular Dynamics Simulations for Drug BindingPreprint
Wu, F., Zhang, Q., Radev, D., Wang, Y., Jin, X., Jiang, Y., Li, S. Z., & Niu, Z. (2022) · Research Square · doi:10.21203/rs.3.rs-1566483/v1 · Cited by 2
Exploring protein sequence similarity with Protein Language UMAPs (PLUMAPs)Preprint
Jinich, A., Nazia, S. Z., & Rhee, K. Y. (2022) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2022.09.27.509824
“Sequence similarity networks (SSNs) have proved to be extremely helpful for exploring homology relationships between proteins 3–6 .”
BackgroundLim, H., Keskin, Ö., Nussinov, R., & Gürsoy, A. (2022) · Research Square · doi:10.21203/rs.3.rs-2035249/v1
CELL-E: zero-shot sequence-to-image synthesis for protein localization predictionPreprint
Khwaja, E., Song, Y. S., & Huang, B. (2023) · Research Square · doi:10.21203/rs.3.rs-2963881/v1
“We also experimented with other types of protein embeddings, including UniRep [28] and ESM1-b [29].”
BackgroundMethodTAPE-Pero: Using Deep Representation Learning Model to Identify and Localize Peroxisomal Proteins
Sui, J., Chen, Y., Cao, Y., & Zhao, Y. (2023) · Lecture notes in computer science · doi:10.1007/978-981-99-4749-2_48
Transmembrane Domain Prediction
Savojardo, C., Manfredi, M., Martelli, P. L., & Casadio, R. (2024) · Elsevier eBooks · doi:10.1016/b978-0-323-95502-7.00075-0
Kalantar, M., Kalanther, I., Kumar, S., Buxton, E. K., & Raeeszadeh-Sarmazdeh, M. (2024) · bioRxiv · doi:10.1101/2024.06.04.597476
“These developments have led to the creation of Large Protein Language Models (LPLMs), which assist in discovering the evolutionary, structural, and functional properties across protein space by encoding amino-acid sequences into numeric vector representations 32 .”
Background
Metadata
- Authors
- Bepler, T., & Berger, B.
- Affiliations
- New York Structural Biology Center (US)
- MIT Computer Science and Artificial Intelligence Laboratory (US)
- Corresponding author
- Bepler, Tristan
- Berger, Bonnie
- Venue
- Cell Systems
- Publisher
- Elsevier BV
- Published
- 2021-06
- Volume
- 12
- Issue
- 6
- Pages
- 654-669.e3
- ISSN
- 2405-4712 (ISSN-L)
- PMC
- PMC8238390
- OpenAlex
- W3166142427
- Cited by
- 483
- References
- 133
- Citations by year
- 2026: 87 · 2025: 120 · 2024: 127 · 2023: 82 · 2022: 55 · 2021: 12
- FWCI
- 21.74
- Type
- journal article
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- Locations
- Cell Systems (publishedVersion, cc-by-nc-nd)
- PubMed (publishedVersion)
- PubMed Central (acceptedVersion)
- DSpace@MIT (Massachusetts Institute of Technology) (submittedVersion, cc-by-nc-nd)
- Funders
- NIH (R35 GM141861)
- Simons Foundation (SF349247)
- Topics
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Machine Learning in Bioinformatics
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Protein Structure and Dynamics
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Fractal and DNA sequence analysis
- Keywords
- protein language models, protein sequence representation, protein structure prediction, protein function prediction, knowledge distillation, transfer learning, protein design, therapeutic design
- MeSH terms
- Machine Learning, Amino Acid Sequence, Language, Proteins, Databases, Protein
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 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
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
