Ariane Mora
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Ariane Mora has a total of 131 co-authors. Here are the top co-authors by number of shared publications:
- Frances H. Arnold
Division of Chemistry and Chemical Engineering, California Institute of Technology, Pasadena, California 91125, United States
9 shared publications - Mikael Bodén
The University of Queensland
8 shared publications - Julia C. Reisenbauer
Division of Chemistry and Chemical Engineering California Institute of Technology Pasadena California USA
5 shared publications - William Jf Rieger
School of Chemistry and Molecular Biology, University of Queensland
4 shared publications - Ana Jaramillo
Division of Chemistry and Chemical Engineering California Institute of Technology Pasadena California USA
2 shared publications Annika Starkenberg
Department of Clinical and Experimental Medicine, Linköping University, SE-58185 Linköping, Sweden
2 shared publications
Latest Publications
- Open Access
Designing enzymes for new-to-nature chemistry and non-natural substrates with AlphaProtein Novo
Creating highly active enzymes for arbitrary reactions is a transformative goal for the molecular sciences 1 . Traditional protein engineering 2,3 is constrained by a reliance on existing natural starting points, which are often challenging to discover. De novo design offers the potential to overcome this by creating enzymes from first principles 4 , but has yet to achieve practically relevant catalytic properties. Here we present AlphaProtein Novo (AP Novo), a machine-learning pipeline that demonstrates, for the first time, that de novo enzyme design can outperform natural sequence mining in addressing challenging chemistry. We used AP Novo to design new-to-nature nitrene transferases to synthesise the pharmacophore piperidine with unprecedented product selectivity, and to create enzymes that degrade the environmental toxin DEHP under conditions that denature natural enzymes. We also obtained state-of-the-art catalytic efficiencies on two well-studied model reactions, and through iterative design and analysis found that key enablers of success were mechanism-inspired metrics based on AlphaFold 3 5 predictions and sequence-ensemble-based scoring of designed backbones. Our results show that de novo design has become a powerful complement to natural enzyme diversity for discovering catalysts for a range of applications.
Open AccessDesigning enzymes for new-to-nature chemistry and non-natural substrates with AlphaProtein Novo
Creating highly active enzymes for arbitrary reactions is a transformative goal for the molecular sciences 1 . Traditional protein engineering 2,3 is constrained by a reliance on existing natural starting points, which are often challenging to discover. De novo design offers the potential to overcome this by creating enzymes from first principles 4 , but has yet to achieve practically relevant catalytic properties. Here we present AlphaProtein Novo (AP Novo), a machine-learning pipeline that demonstrates, for the first time, that de novo enzyme design can outperform natural sequence mining in addressing challenging chemistry. We used AP Novo to design new-to-nature nitrene transferases to synthesise the pharmacophore piperidine with unprecedented product selectivity, and to create enzymes that degrade the environmental toxin DEHP under conditions that denature natural enzymes. We also obtained state-of-the-art catalytic efficiencies on two well-studied model reactions, and through iterative design and analysis found that key enablers of success were mechanism-inspired metrics based on AlphaFold 3 5 predictions and sequence-ensemble-based scoring of designed backbones. Our results show that de novo design has become a powerful complement to natural enzyme diversity for discovering catalysts for a range of applications.
- Open Access
Reframing enzyme function prediction as conditional generation
Enzymes frequently exhibit promiscuous activity beyond their native roles, providing starting-points for new functions. Finding these promiscuous enzymes, especially for non-native chemical transformations, is challenging but highly valuable, as they promise novel, sustainable solutions for chemistry and biotechnology. However, current machine learning methods are poorly suited to discovering unseen chemistry as they often frame function prediction as closed set classification or a retrieval task. Here, we present Fluxion , a generative deep learning framework that learns enzymatic catalysis by modeling dynamic electron flow trajectories across the enzyme’s catalytic residues. By combining both synthetic chemistry and biochemical datasets with protein language model representations, Fluxion generates multi-step electron-flow trajectories analogous to the arrow-pushing representations used to describe enzyme reaction mechanisms. Generation is conditioned on enzyme context, including the enzyme sequence, catalytic residues, substrates, and cofactors. We show that this conditioning allows Fluxion to learn enzyme-dependent regioselectivity across cytochrome P450 enzymes with different sequences shifting the predicted reaction sites for the same substrate. We then demonstrate that Fluxion’s embeddings are useful for downstream tasks, such as specificity prediction on two experimental datasets, with and without finetuning. Finally, we show that Fluxion has the potential to transfer synthetic chemical logic to biology; it can generate the observed non-native product from real-world non-native directed evolution screens. Our results establish a proof of concept that generative modeling through mechanistic representations of enzymes can shift enzyme function prediction beyond static database retrieval and closed set classification to function generation. This conceptual framework provides a stepping stone towards an in silico generative method to discover non-native biocatalysts.
Open AccessReframing enzyme function prediction as conditional generation
Enzymes frequently exhibit promiscuous activity beyond their native roles, providing starting-points for new functions. Finding these promiscuous enzymes, especially for non-native chemical transformations, is challenging but highly valuable, as they promise novel, sustainable solutions for chemistry and biotechnology. However, current machine learning methods are poorly suited to discovering unseen chemistry as they often frame function prediction as closed set classification or a retrieval task. Here, we present Fluxion , a generative deep learning framework that learns enzymatic catalysis by modeling dynamic electron flow trajectories across the enzyme’s catalytic residues. By combining both synthetic chemistry and biochemical datasets with protein language model representations, Fluxion generates multi-step electron-flow trajectories analogous to the arrow-pushing representations used to describe enzyme reaction mechanisms. Generation is conditioned on enzyme context, including the enzyme sequence, catalytic residues, substrates, and cofactors. We show that this conditioning allows Fluxion to learn enzyme-dependent regioselectivity across cytochrome P450 enzymes with different sequences shifting the predicted reaction sites for the same substrate. We then demonstrate that Fluxion’s embeddings are useful for downstream tasks, such as specificity prediction on two experimental datasets, with and without finetuning. Finally, we show that Fluxion has the potential to transfer synthetic chemical logic to biology; it can generate the observed non-native product from real-world non-native directed evolution screens. Our results establish a proof of concept that generative modeling through mechanistic representations of enzymes can shift enzyme function prediction beyond static database retrieval and closed set classification to function generation. This conceptual framework provides a stepping stone towards an in silico generative method to discover non-native biocatalysts.
- Open Access
- 4 September 2026
- Published by eLife Sciences Publications, Ltd in eLife
- Vol. 14
Squidly harnesses enzyme functional hierarchy and contrastive learning to efficiently predict catalytic residues from sequence
Enzymes present a sustainable alternative to traditional chemical industries, drug synthesis, and bioremediation applications. Because catalytic residues are the key amino acids that drive enzyme function, their accurate prediction facilitates enzyme function prediction. Sequence similarity-based approaches such as BLAST are fast but require previously annotated homologues. Machine-learning (ML) approaches aim to overcome this limitation; however, current gold-standard ML-based methods require high-quality 3D structures limiting their application to large datasets. To address these challenges, we developed Squidly, a sequence-only tool that leverages contrastive representation learning with a biology-informed, rationally designed pairing scheme to distinguish catalytic from non-catalytic residues using per-token Protein Language Model embeddings. Squidly surpasses state-of-the-art ML annotation methods in catalytic residue prediction while remaining sufficiently fast to enable wide-scale screening of databases. We ensemble Squidly with BLAST to provide an efficient tool that annotates catalytic residues with high precision and recall for both in- and out-of-distribution sequences.
- 4 September 2026
- Published by eLife Sciences Publications, Ltd in eLife
- Vol. 14
Open AccessSquidly harnesses enzyme functional hierarchy and contrastive learning to efficiently predict catalytic residues from sequence
Enzymes present a sustainable alternative to traditional chemical industries, drug synthesis, and bioremediation applications. Because catalytic residues are the key amino acids that drive enzyme function, their accurate prediction facilitates enzyme function prediction. Sequence similarity-based approaches such as BLAST are fast but require previously annotated homologues. Machine-learning (ML) approaches aim to overcome this limitation; however, current gold-standard ML-based methods require high-quality 3D structures limiting their application to large datasets. To address these challenges, we developed Squidly, a sequence-only tool that leverages contrastive representation learning with a biology-informed, rationally designed pairing scheme to distinguish catalytic from non-catalytic residues using per-token Protein Language Model embeddings. Squidly surpasses state-of-the-art ML annotation methods in catalytic residue prediction while remaining sufficiently fast to enable wide-scale screening of databases. We ensemble Squidly with BLAST to provide an efficient tool that annotates catalytic residues with high precision and recall for both in- and out-of-distribution sequences.
- Open Access
- 27 July 2026
- Published by National Academy of Sciences in Proceedings of the National Academy of Sciences
- Vol. 123 (31)
Discovery of a phenazine–thiol conjugase from sparse data using genome-informed machine learning
Machine learning has enabled powerful biological discoveries using models trained on large datasets. However, for many important biological questions, such as identifying enzymes that transform understudied substrates, sparsity of training data is often a major bottleneck. Here, using phenazine natural products as a case study, we show that integrating genome-informed data augmentation with contrastive learning in protein language space enables identification of phenazine-interacting proteins starting from only 14 known phenazine modifying sequences. We name this approach ML-CITO (Machine Learning for genomic Context-Informed Transferable discOvery). Applying this framework led to the discovery of PTC (phenazine–thiol conjugase), an enzyme known to catalyze phenazine thioconjugation, a phenazine modification reaction long observed but previously presumed to occur only through nonenzymatic chemistry. In silico simulation and experimental measurements demonstrate that PTC binds to both phenazine and glutathione as substrates. Recombinant expression and biochemical characterization reveal that PTC promotes glutathione-dependent modification of phenazines, yielding distinct reaction outcomes that depend on substrate identity. Although thiol-conjugated phenazine products exhibit reduced toxicity to bacterial cells, deletion of the gene encoding PTC does not confer a strong fitness disadvantage, illustrating how direct learning of sequences can uncover relevant enzymes that might evade phenotype-based genetic screens. Together, these results demonstrate that coupling comparative genomics with protein machine learning can convert “small data” typically outside the scope of machine learning into actionable predictive power, thereby facilitating enzyme discovery.
- 27 July 2026
- Published by National Academy of Sciences in Proceedings of the National Academy of Sciences
- Vol. 123 (31)
Open AccessDiscovery of a phenazine–thiol conjugase from sparse data using genome-informed machine learning
Machine learning has enabled powerful biological discoveries using models trained on large datasets. However, for many important biological questions, such as identifying enzymes that transform understudied substrates, sparsity of training data is often a major bottleneck. Here, using phenazine natural products as a case study, we show that integrating genome-informed data augmentation with contrastive learning in protein language space enables identification of phenazine-interacting proteins starting from only 14 known phenazine modifying sequences. We name this approach ML-CITO (Machine Learning for genomic Context-Informed Transferable discOvery). Applying this framework led to the discovery of PTC (phenazine–thiol conjugase), an enzyme known to catalyze phenazine thioconjugation, a phenazine modification reaction long observed but previously presumed to occur only through nonenzymatic chemistry. In silico simulation and experimental measurements demonstrate that PTC binds to both phenazine and glutathione as substrates. Recombinant expression and biochemical characterization reveal that PTC promotes glutathione-dependent modification of phenazines, yielding distinct reaction outcomes that depend on substrate identity. Although thiol-conjugated phenazine products exhibit reduced toxicity to bacterial cells, deletion of the gene encoding PTC does not confer a strong fitness disadvantage, illustrating how direct learning of sequences can uncover relevant enzymes that might evade phenotype-based genetic screens. Together, these results demonstrate that coupling comparative genomics with protein machine learning can convert “small data” typically outside the scope of machine learning into actionable predictive power, thereby facilitating enzyme discovery.
- Open Access
Discovery of a phenazine–thiol conjugase from sparse data using genome-informed machine learning
Machine learning has enabled powerful biological discoveries using models trained on large datasets. However, for many important biological questions, such as identifying enzymes that transform understudied substrates, sparsity of training data is often a major bottleneck. Here, using phenazine natural products as a case study, we show that integrating genome-informed data augmentation with contrastive learning in protein language space enables identification of phenazine-interacting proteins starting from only 14 known phenazine modifying sequences. Applying this framework led to the discovery of PTC (Phenazine-Thiol Conjugase), the first enzyme known to catalyze phenazine thioconjugation, a phenazine modification reaction long observed but previously presumed to occur only through non-enzymatic chemistry. In silico simulation and experimental measurements demonstrate that PTC binds to both phenazine and glutathione as substrates. Recombinant expression and biochemical characterization reveal that PTC promotes glutathione-dependent modification of phenazines, yielding distinct reaction outcomes that depend on substrate identity. Although thiol-conjugated phenazine products exhibit reduced toxicity to bacterial cells, deletion of the gene encoding PTC does not confer a strong fitness disadvantage, illustrating how direct learning of sequences can uncover relevant enzymes that might evade phenotype-based genetic screens. Together, these results demonstrate that coupling comparative genomics with protein machine learning can convert “small data” typically outside the scope of machine learning into actionable predictive power, thereby facilitating enzyme discovery. Significance: Machine learning excels when large, well-labeled datasets are available, yet many biologically important problems lack sufficient experimental data to support such approaches to discovery. This limitation is particularly acute for identifying enzymes acting on rare or understudied substrates. Here, we show that genomic organization can be leveraged as an additional source of biological information to address data sparsity. Starting with only 14 enzymes experimentally shown to modify phenazines, we developed a model identifying phenazine-interacting enzymes by integrating genome-informed data augmentation with protein machine learning. Guided by the model, we discovered the first enzyme known to catalyze thioconjugation modifications of phenazines, demonstrating a simple yet powerful strategy for extracting predictive insight from sparse biological knowledge.
Open AccessDiscovery of a phenazine–thiol conjugase from sparse data using genome-informed machine learning
Machine learning has enabled powerful biological discoveries using models trained on large datasets. However, for many important biological questions, such as identifying enzymes that transform understudied substrates, sparsity of training data is often a major bottleneck. Here, using phenazine natural products as a case study, we show that integrating genome-informed data augmentation with contrastive learning in protein language space enables identification of phenazine-interacting proteins starting from only 14 known phenazine modifying sequences. Applying this framework led to the discovery of PTC (Phenazine-Thiol Conjugase), the first enzyme known to catalyze phenazine thioconjugation, a phenazine modification reaction long observed but previously presumed to occur only through non-enzymatic chemistry. In silico simulation and experimental measurements demonstrate that PTC binds to both phenazine and glutathione as substrates. Recombinant expression and biochemical characterization reveal that PTC promotes glutathione-dependent modification of phenazines, yielding distinct reaction outcomes that depend on substrate identity. Although thiol-conjugated phenazine products exhibit reduced toxicity to bacterial cells, deletion of the gene encoding PTC does not confer a strong fitness disadvantage, illustrating how direct learning of sequences can uncover relevant enzymes that might evade phenotype-based genetic screens. Together, these results demonstrate that coupling comparative genomics with protein machine learning can convert “small data” typically outside the scope of machine learning into actionable predictive power, thereby facilitating enzyme discovery. Significance: Machine learning excels when large, well-labeled datasets are available, yet many biologically important problems lack sufficient experimental data to support such approaches to discovery. This limitation is particularly acute for identifying enzymes acting on rare or understudied substrates. Here, we show that genomic organization can be leveraged as an additional source of biological information to address data sparsity. Starting with only 14 enzymes experimentally shown to modify phenazines, we developed a model identifying phenazine-interacting enzymes by integrating genome-informed data augmentation with protein machine learning. Guided by the model, we discovered the first enzyme known to catalyze thioconjugation modifications of phenazines, demonstrating a simple yet powerful strategy for extracting predictive insight from sparse biological knowledge.
- 24 February 2026
- Published by Wiley in Angewandte Chemie
- Vol. 138 (14)
A Chemoselective and Stereodivergent Platform of Heme‐Nitrene Transferases to Access Chiral Aryl‐β‐Amino Esters and An Investigation of the Sequence‐Activity Landscape
Engineered biocatalysts can utilize nitrene precursors to access enantioenriched amination products, yet they have not been applied to produce valuable, enantiomerically enriched noncanonical β‐amino esters. Current approaches to synthesizing β‐amino acids rely on pre‐oxidized precursors and multistep synthetic approaches involving various protecting groups. We engineered a platform of heme enzymes for stereoselective C–H bond amination of readily available carboxylic ester derivatives to install primary amines. A directed evolution campaign coupled with sequencing of over 1000 variants enabled us to develop engineered variants that use either O ‐pivaloylhydroxylamine triflic acid (PONT) or hydroxylamine hydrochloride (H 2 NOH∙HCl) as aminating reagents. An analysis of the resulting sequence–activity dataset revealed additional improvements that could be made to the final variant, highlighting the utility of sequencing data to guide future steps in directed evolution campaigns. The evolved nitrene transferases expand the scope of accessible chiral β‐amino acid building blocks for peptidomimetic applications and provide new starting points for the design and synthesis of enantioenriched β‐amino acid motifs.
- 24 February 2026
- Published by Wiley in Angewandte Chemie
- Vol. 138 (14)
A Chemoselective and Stereodivergent Platform of Heme‐Nitrene Transferases to Access Chiral Aryl‐β‐Amino Esters and An Investigation of the Sequence‐Activity Landscape
Engineered biocatalysts can utilize nitrene precursors to access enantioenriched amination products, yet they have not been applied to produce valuable, enantiomerically enriched noncanonical β‐amino esters. Current approaches to synthesizing β‐amino acids rely on pre‐oxidized precursors and multistep synthetic approaches involving various protecting groups. We engineered a platform of heme enzymes for stereoselective C–H bond amination of readily available carboxylic ester derivatives to install primary amines. A directed evolution campaign coupled with sequencing of over 1000 variants enabled us to develop engineered variants that use either O ‐pivaloylhydroxylamine triflic acid (PONT) or hydroxylamine hydrochloride (H 2 NOH∙HCl) as aminating reagents. An analysis of the resulting sequence–activity dataset revealed additional improvements that could be made to the final variant, highlighting the utility of sequencing data to guide future steps in directed evolution campaigns. The evolved nitrene transferases expand the scope of accessible chiral β‐amino acid building blocks for peptidomimetic applications and provide new starting points for the design and synthesis of enantioenriched β‐amino acid motifs.
- 17 February 2026
- Published by Wiley in Angewandte Chemie International Edition
- Vol. 65 (14) , e25505
A Chemoselective and Stereodivergent Platform of Heme‐Nitrene Transferases to Access Chiral Aryl‐β‐Amino Esters and An Investigation of the Sequence‐Activity Landscape
Engineered biocatalysts can utilize nitrene precursors to access enantioenriched amination products, yet they have not been applied to produce valuable, enantiomerically enriched noncanonical β‐amino esters. Current approaches to synthesizing β‐amino acids rely on pre‐oxidized precursors and multistep synthetic approaches involving various protecting groups. We engineered a platform of heme enzymes for stereoselective C–H bond amination of readily available carboxylic ester derivatives to install primary amines. A directed evolution campaign coupled with sequencing of over 1000 variants enabled us to develop engineered variants that use either O ‐pivaloylhydroxylamine triflic acid (PONT) or hydroxylamine hydrochloride (H 2 NOH∙HCl) as aminating reagents. An analysis of the resulting sequence–activity dataset revealed additional improvements that could be made to the final variant, highlighting the utility of sequencing data to guide future steps in directed evolution campaigns. The evolved nitrene transferases expand the scope of accessible chiral β‐amino acid building blocks for peptidomimetic applications and provide new starting points for the design and synthesis of enantioenriched β‐amino acid motifs.
- 17 February 2026
- Published by Wiley in Angewandte Chemie International Edition
- Vol. 65 (14) , e25505
A Chemoselective and Stereodivergent Platform of Heme‐Nitrene Transferases to Access Chiral Aryl‐β‐Amino Esters and An Investigation of the Sequence‐Activity Landscape
Engineered biocatalysts can utilize nitrene precursors to access enantioenriched amination products, yet they have not been applied to produce valuable, enantiomerically enriched noncanonical β‐amino esters. Current approaches to synthesizing β‐amino acids rely on pre‐oxidized precursors and multistep synthetic approaches involving various protecting groups. We engineered a platform of heme enzymes for stereoselective C–H bond amination of readily available carboxylic ester derivatives to install primary amines. A directed evolution campaign coupled with sequencing of over 1000 variants enabled us to develop engineered variants that use either O ‐pivaloylhydroxylamine triflic acid (PONT) or hydroxylamine hydrochloride (H 2 NOH∙HCl) as aminating reagents. An analysis of the resulting sequence–activity dataset revealed additional improvements that could be made to the final variant, highlighting the utility of sequencing data to guide future steps in directed evolution campaigns. The evolved nitrene transferases expand the scope of accessible chiral β‐amino acid building blocks for peptidomimetic applications and provide new starting points for the design and synthesis of enantioenriched β‐amino acid motifs.
- 10 February 2026
- Published by University of Queensland Library
Machine learning models of epigenetic dynamics driving cell fate
- 10 February 2026
- Published by University of Queensland Library
Machine learning models of epigenetic dynamics driving cell fate
- Open Access
A unified pipeline for discovering previously unknown enzyme activities
Enzymes catalyze diverse chemical transformations and offer a sustainable approach to both breaking and making chemical bonds. However, finding an enzyme capable of performing a specific chemical reaction remains a challenge. We developed a new framework, Enzyme-toolkit (Enzyme-tk), that integrates 23 open-source tools to enable the discovery of enzymes that have activity toward a specific target reaction. Additionally, we introduce two new methods to facilitate enzyme discovery: (1) Func-e, an ML tool that searches large databases for enzymes that potentially catalyze a specific chemical transformation and (2) Oligopoolio, a gene assembly approach that reduces the cost of accessing protein sequences and thus the barrier to their experimental validation. We applied Enzyme-tk to find enzymes for chemical degradation of two man-made pollutants, di-(2-ethylhexyl) phthalate (DEHP) and triphenyl phosphate (TPP). We demonstrate that new, previously unannotated enzymes with favorable characteristics, such as high thermostability, can be identified using Enzyme-tk for reactions that are dissimilar to the training set.
Open AccessA unified pipeline for discovering previously unknown enzyme activities
Enzymes catalyze diverse chemical transformations and offer a sustainable approach to both breaking and making chemical bonds. However, finding an enzyme capable of performing a specific chemical reaction remains a challenge. We developed a new framework, Enzyme-toolkit (Enzyme-tk), that integrates 23 open-source tools to enable the discovery of enzymes that have activity toward a specific target reaction. Additionally, we introduce two new methods to facilitate enzyme discovery: (1) Func-e, an ML tool that searches large databases for enzymes that potentially catalyze a specific chemical transformation and (2) Oligopoolio, a gene assembly approach that reduces the cost of accessing protein sequences and thus the barrier to their experimental validation. We applied Enzyme-tk to find enzymes for chemical degradation of two man-made pollutants, di-(2-ethylhexyl) phthalate (DEHP) and triphenyl phosphate (TPP). We demonstrate that new, previously unannotated enzymes with favorable characteristics, such as high thermostability, can be identified using Enzyme-tk for reactions that are dissimilar to the training set.
- Open Access
- 2 January 2026
- Published by eLife Sciences Publications, Ltd in eLife preprints
Squidly: Enzyme Catalytic Residue Prediction Harnessing a Biology-Informed Contrastive Learning Framework
Enzymes present a sustainable alternative to traditional chemical industries, drug synthesis, and bioremediation applications. Because catalytic residues are the key amino acids that drive enzyme function, their accurate prediction facilitates enzyme function prediction. Sequence similarity-based approaches such as BLAST are fast but require previously annotated homologs. Machine learning approaches aim to overcome this limitation; however, current gold-standard machine learning (ML)-based methods require high-quality 3D structures limiting their application to large datasets. To address these challenges, we developed Squidly, a sequence-only tool that leverages contrastive representation learning with a biology-informed, rationally designed pairing scheme to distinguish catalytic from non-catalytic residues using per-token Protein Language Model embeddings. Squidly surpasses state-of-the-art ML annotation methods in catalytic residue prediction while remaining sufficiently fast to enable wide-scale screening of databases. We ensemble Squidly with BLAST to provide an efficient tool that annotates catalytic residues with high precision and recall for both in- and out-of-distribution sequences.
- 2 January 2026
- Published by eLife Sciences Publications, Ltd in eLife preprints
Open AccessSquidly: Enzyme Catalytic Residue Prediction Harnessing a Biology-Informed Contrastive Learning Framework
Enzymes present a sustainable alternative to traditional chemical industries, drug synthesis, and bioremediation applications. Because catalytic residues are the key amino acids that drive enzyme function, their accurate prediction facilitates enzyme function prediction. Sequence similarity-based approaches such as BLAST are fast but require previously annotated homologs. Machine learning approaches aim to overcome this limitation; however, current gold-standard machine learning (ML)-based methods require high-quality 3D structures limiting their application to large datasets. To address these challenges, we developed Squidly, a sequence-only tool that leverages contrastive representation learning with a biology-informed, rationally designed pairing scheme to distinguish catalytic from non-catalytic residues using per-token Protein Language Model embeddings. Squidly surpasses state-of-the-art ML annotation methods in catalytic residue prediction while remaining sufficiently fast to enable wide-scale screening of databases. We ensemble Squidly with BLAST to provide an efficient tool that annotates catalytic residues with high precision and recall for both in- and out-of-distribution sequences.