Anette Henriksen
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Anette Henriksen has a total of 163 co-authors. Here are the top co-authors by number of shared publications:
- Michael Gajhede
University of Copenhagen
19 shared publications - Birte Svensson
Enzyme and Protein Chemistry, Department of Biotechnology and Biomedicine Technical University of Denmark Kgs. Lyngby Denmark
12 shared publications - Karen Gjesing Welinder
Department of Biotechnology, Chemistry and Environmental Engineering, Aalborg University, Sohngaardsholmsvej 49, DK-9000 Aalborg, Denmark
7 shared publications - Osman Mirza
Department of Drug Design and Pharmacology, Faculty of Health and Medical Sciences, University of Copenhagen, Universitetsparken 2, DK-2100 Copenhagen, Denmark
7 shared publications - Per Hägglund
Department of Biomedical Sciences, University of Copenhagen, Panum, Blegdamsvej 3B, Copenhagen N, DK-2200, Denmark
7 shared publications - Christine Finnie
Technical University of Denmark
6 shared publications
Latest Publications
- Open Access
- 27 May 2026
- Published by Taylor & Francis in mAbs
- Vol. 18 (1) , 2678000
Prediction of antibody non-specificity using protein language models and biophysical parameters
The development of therapeutic antibodies requires optimizing target binding affinity and pharmacodynamics, while ensuring high developability potential, including minimizing non-specific binding. In this study, we address this problem by predicting antibody non-specificity by two complementary approaches: (1) antibody sequence embeddings by protein language models (PLMs) and (2) a comprehensive set of sequence-based biophysical descriptors. We benchmark the PLM embeddings against interpretable, sequence-derived biophysical descriptors and use fragment-specific models (variable heavy (VH) and variable light (VL) chains, concatenated regions, and individual complementary-determining regions (CDRs) to identify region-level determinants of non-specificity in functional antibodies that bind defined targets. These models were trained on previously published human and mouse antibody data and tested on three public datasets. We show that non-specificity is best predicted from the VH domain and heavy-chain CDRs. These region-level analyses show that heavy-chain features, especially H-CDR3, dominate the predictive signal. The top performing PLM, a VH domain-based Evolutionary Scale Modeling 1 v LogisticReg model, resulted in 10-fold cross-validation accuracy of up to 71%. While predictive accuracy is comparable to classical machine-learning baselines, PLM-based embeddings provide sequence-context representations that yield consistent region-level attribution and provide complementarity to the classical models by enhancing the reliability of predicted non‑specificity scores when used in combination. Our biophysical descriptor-based analysis identified the isoelectric point as a key driver of non-specificity, consistent with previous reports. Our findings highlight the importance of biophysical properties in predicting antibody non-specificity and highlight the potential of PLMs for the development of antibody-based therapeutics. These conclusions are robust to alternative class definitions and are not driven by VH-VL mutational bias. We illustrate the generalizability and practical use of the PLM approach by extending it to therapeutic antibodies and nanobodies, providing a tool for early-stage developability assessment, and we make the models publicly available.
- 27 May 2026
- Published by Taylor & Francis in mAbs
- Vol. 18 (1) , 2678000
Open AccessPrediction of antibody non-specificity using protein language models and biophysical parameters
The development of therapeutic antibodies requires optimizing target binding affinity and pharmacodynamics, while ensuring high developability potential, including minimizing non-specific binding. In this study, we address this problem by predicting antibody non-specificity by two complementary approaches: (1) antibody sequence embeddings by protein language models (PLMs) and (2) a comprehensive set of sequence-based biophysical descriptors. We benchmark the PLM embeddings against interpretable, sequence-derived biophysical descriptors and use fragment-specific models (variable heavy (VH) and variable light (VL) chains, concatenated regions, and individual complementary-determining regions (CDRs) to identify region-level determinants of non-specificity in functional antibodies that bind defined targets. These models were trained on previously published human and mouse antibody data and tested on three public datasets. We show that non-specificity is best predicted from the VH domain and heavy-chain CDRs. These region-level analyses show that heavy-chain features, especially H-CDR3, dominate the predictive signal. The top performing PLM, a VH domain-based Evolutionary Scale Modeling 1 v LogisticReg model, resulted in 10-fold cross-validation accuracy of up to 71%. While predictive accuracy is comparable to classical machine-learning baselines, PLM-based embeddings provide sequence-context representations that yield consistent region-level attribution and provide complementarity to the classical models by enhancing the reliability of predicted non‑specificity scores when used in combination. Our biophysical descriptor-based analysis identified the isoelectric point as a key driver of non-specificity, consistent with previous reports. Our findings highlight the importance of biophysical properties in predicting antibody non-specificity and highlight the potential of PLMs for the development of antibody-based therapeutics. These conclusions are robust to alternative class definitions and are not driven by VH-VL mutational bias. We illustrate the generalizability and practical use of the PLM approach by extending it to therapeutic antibodies and nanobodies, providing a tool for early-stage developability assessment, and we make the models publicly available.
- Open Access
Prediction of Antibody Non-Specificity using Protein Language Models and Biophysical Parameters
The development of therapeutic antibodies requires optimizing target binding affinity and pharmacodynamics, while ensuring high developability potential, including minimizing non-specific binding. In this study, we address this problem by predicting antibody non-specificity by two complementary approaches: (i) antibody sequence embeddings by protein language models (PLMs), and (ii) a comprehensive set of sequence-based biophysical descriptors. These models were trained on human and mouse antibody data from Boughter et al . (2020) and tested on three public datasets: Jain et al . (2017), Shehata et al . (2019) and Harvey et al . (2022). We show that non-specificity is best predicted from the heavy variable domain and heavy-chain complementary variable regions (CDRs). The top performing PLM, a heavy variable domain-based ESM 1v LogisticReg model, resulted in 10-fold cross-validation accuracy of up to 71%. Our biophysical descriptor-based analysis identified the isoelectric point as a key driver of non-specificity. Our findings underscore the importance of biophysical properties in predicting antibody non-specificity and highlight the potential of protein language models for the development of antibody-based therapeutics. To illustrate the use of our approach in the development of lead candidates with high developability potential, we show that it can be extended to therapeutic antibodies and nanobodies.
Open AccessPrediction of Antibody Non-Specificity using Protein Language Models and Biophysical Parameters
The development of therapeutic antibodies requires optimizing target binding affinity and pharmacodynamics, while ensuring high developability potential, including minimizing non-specific binding. In this study, we address this problem by predicting antibody non-specificity by two complementary approaches: (i) antibody sequence embeddings by protein language models (PLMs), and (ii) a comprehensive set of sequence-based biophysical descriptors. These models were trained on human and mouse antibody data from Boughter et al . (2020) and tested on three public datasets: Jain et al . (2017), Shehata et al . (2019) and Harvey et al . (2022). We show that non-specificity is best predicted from the heavy variable domain and heavy-chain complementary variable regions (CDRs). The top performing PLM, a heavy variable domain-based ESM 1v LogisticReg model, resulted in 10-fold cross-validation accuracy of up to 71%. Our biophysical descriptor-based analysis identified the isoelectric point as a key driver of non-specificity. Our findings underscore the importance of biophysical properties in predicting antibody non-specificity and highlight the potential of protein language models for the development of antibody-based therapeutics. To illustrate the use of our approach in the development of lead candidates with high developability potential, we show that it can be extended to therapeutic antibodies and nanobodies.
- Open Access
- 3 May 2018
- Published by Wiley in Haemophilia
- Vol. 24 (6) , 880-887
A comparative analysis of heterogeneity in commercially available recombinant factor VIII products
Introduction: Advances in analytical technologies enable investigation of possible correlations between molecular structure, aggregation and subvisible particle content. Regulatory agencies place increasing attention on potential risks associated with protein aggregates in the micron range in biological therapeutics.Aim: Assess the heterogeneity, high‐molecular‐weight protein (HMWP) species, subvisible particle content and posttranslational modifications in six commercially available recombinant FVIII (rFVIII) products.Methods: Three B‐domain‐deleted (BDD) or B‐domain truncated rFVIII products (turoctocog alfa, simoctocog alfa and moroctocog alfa) and three full‐length rFVIII products (octocog alfa FS and two octocog alfa) were analysed. HMWP content, amount of micron range subvisible particles, tyrosine‐1680 sulphation and N‐glycan analysis were investigated.Results: The B‐domain‐modified products had more protein size homogeneity vs the full‐length products. Size exclusion‐high‐performance liquid chromatography data indicated no association between B‐domain structure and aggregate content or size of the products tested. The rFVIII products showed large variation in subvisible particle concentration, with turoctocog alfa and simoctocog alfa having the lowest numbers (1000‐1600 and 1800‐2400 particles/100 IU, respectively). Turoctocog alfa and simoctocog alfa displayed the most complete tyrosine sulphation (>99.5%).Conclusion: Overall, there was no association between molecular structure (full‐length B‐domain, BDD or truncated) and subvisible particle or HMWP content. Dissimilarities may be related to production and product handling differences. In this study, turoctocog alfa, such as simoctocog alfa, had one of the lowest levels of subvisible particles and HMWP content, and high protein size homogeneity.
- 3 May 2018
- Published by Wiley in Haemophilia
- Vol. 24 (6) , 880-887
Open AccessA comparative analysis of heterogeneity in commercially available recombinant factor VIII products
Introduction: Advances in analytical technologies enable investigation of possible correlations between molecular structure, aggregation and subvisible particle content. Regulatory agencies place increasing attention on potential risks associated with protein aggregates in the micron range in biological therapeutics.Aim: Assess the heterogeneity, high‐molecular‐weight protein (HMWP) species, subvisible particle content and posttranslational modifications in six commercially available recombinant FVIII (rFVIII) products.Methods: Three B‐domain‐deleted (BDD) or B‐domain truncated rFVIII products (turoctocog alfa, simoctocog alfa and moroctocog alfa) and three full‐length rFVIII products (octocog alfa FS and two octocog alfa) were analysed. HMWP content, amount of micron range subvisible particles, tyrosine‐1680 sulphation and N‐glycan analysis were investigated.Results: The B‐domain‐modified products had more protein size homogeneity vs the full‐length products. Size exclusion‐high‐performance liquid chromatography data indicated no association between B‐domain structure and aggregate content or size of the products tested. The rFVIII products showed large variation in subvisible particle concentration, with turoctocog alfa and simoctocog alfa having the lowest numbers (1000‐1600 and 1800‐2400 particles/100 IU, respectively). Turoctocog alfa and simoctocog alfa displayed the most complete tyrosine sulphation (>99.5%).Conclusion: Overall, there was no association between molecular structure (full‐length B‐domain, BDD or truncated) and subvisible particle or HMWP content. Dissimilarities may be related to production and product handling differences. In this study, turoctocog alfa, such as simoctocog alfa, had one of the lowest levels of subvisible particles and HMWP content, and high protein size homogeneity.
- 1 November 2016
- Published by Elsevier in Journal of Pharmaceutical Sciences
- Vol. 105 (11) , 3366-3375
A Comprehensive Evaluation of Nanoparticle Tracking Analysis (NanoSight) for Characterization of Proteinaceous Submicron Particles
- 1 November 2016
- Published by Elsevier in Journal of Pharmaceutical Sciences
- Vol. 105 (11) , 3366-3375
A Comprehensive Evaluation of Nanoparticle Tracking Analysis (NanoSight) for Characterization of Proteinaceous Submicron Particles
- 1 August 2016
- Published by Elsevier in Biochimica et Biophysica Acta (BBA) - Proteins and Proteomics
- Vol. 1864 (8) , 974-982
Seed thioredoxin h
- 1 August 2016
- Published by Elsevier in Biochimica et Biophysica Acta (BBA) - Proteins and Proteomics
- Vol. 1864 (8) , 974-982
Seed thioredoxin h
- 2 May 2016
- Published by Springer Nature in Cellular and Molecular Life Sciences
- Vol. 73 (14) , 2619-2641
Structure and function of α-glucan debranching enzymes
α-Glucan debranching enzymes hydrolyse α-1,6-linkages in starch/glycogen, thereby, playing a central role in energy metabolism in all living organisms. They belong to glycoside hydrolase families GH13 and GH57 and several of these enzymes are industrially important. Nine GH13 subfamilies include α-glucan debranching enzymes; isoamylase and glycogen debranching enzymes (GH13_11); pullulanase type I/limit dextrinase (GH13_12–14); pullulan hydrolase (GH13_20); bifunctional glycogen debranching enzyme (GH13_25); oligo-1 and glucan-1,6-α-glucosidases (GH13_31); pullulanase type II (GH13_39); and α-amylase domains (GH13_41) in two-domain amylase–pullulanases. GH57 harbours type II pullulanases. Specificity differences, domain organisation, carbohydrate binding modules, sequence motifs, three-dimensional structures and specificity determinants are discussed. The phylogenetic analysis indicated that GH13_39 enzymes could represent a “missing link” between the strictly α-1,6-specific debranching enzymes and the enzymes with dual specificity and α-1,4-linkage preference.
- 2 May 2016
- Published by Springer Nature in Cellular and Molecular Life Sciences
- Vol. 73 (14) , 2619-2641
Structure and function of α-glucan debranching enzymes
α-Glucan debranching enzymes hydrolyse α-1,6-linkages in starch/glycogen, thereby, playing a central role in energy metabolism in all living organisms. They belong to glycoside hydrolase families GH13 and GH57 and several of these enzymes are industrially important. Nine GH13 subfamilies include α-glucan debranching enzymes; isoamylase and glycogen debranching enzymes (GH13_11); pullulanase type I/limit dextrinase (GH13_12–14); pullulan hydrolase (GH13_20); bifunctional glycogen debranching enzyme (GH13_25); oligo-1 and glucan-1,6-α-glucosidases (GH13_31); pullulanase type II (GH13_39); and α-amylase domains (GH13_41) in two-domain amylase–pullulanases. GH57 harbours type II pullulanases. Specificity differences, domain organisation, carbohydrate binding modules, sequence motifs, three-dimensional structures and specificity determinants are discussed. The phylogenetic analysis indicated that GH13_39 enzymes could represent a “missing link” between the strictly α-1,6-specific debranching enzymes and the enzymes with dual specificity and α-1,4-linkage preference.
- Open Access
- 1 May 2015
- Published by Elsevier in Journal of Biological Chemistry
- Vol. 290 (20) , 12614-12629
Crystal Structure of Barley Limit Dextrinase-Limit Dextrinase Inhibitor (LD-LDI) Complex Reveals Insights into Mechanism and Diversity of Cereal Type Inhibitors
- 1 May 2015
- Published by Elsevier in Journal of Biological Chemistry
- Vol. 290 (20) , 12614-12629
Open AccessCrystal Structure of Barley Limit Dextrinase-Limit Dextrinase Inhibitor (LD-LDI) Complex Reveals Insights into Mechanism and Diversity of Cereal Type Inhibitors
- 1 March 2015
- Published by Elsevier in Journal of Molecular Biology
- Vol. 427 (6) , 1263-1277
Oligosaccharide and Substrate Binding in the Starch Debranching Enzyme Barley Limit Dextrinase
- 1 March 2015
- Published by Elsevier in Journal of Molecular Biology
- Vol. 427 (6) , 1263-1277
Oligosaccharide and Substrate Binding in the Starch Debranching Enzyme Barley Limit Dextrinase
- Open Access
- 1 August 2014
- Published by Elsevier in Journal of Biological Chemistry
- Vol. 289 (33) , 22991-23003
Crystal Structure of the Chlamydomonas Starch Debranching Enzyme Isoamylase ISA1 Reveals Insights into the Mechanism of Branch Trimming and Complex Assembly
- 1 August 2014
- Published by Elsevier in Journal of Biological Chemistry
- Vol. 289 (33) , 22991-23003
Open AccessCrystal Structure of the Chlamydomonas Starch Debranching Enzyme Isoamylase ISA1 Reveals Insights into the Mechanism of Branch Trimming and Complex Assembly
- 1 March 2014
- Published by Elsevier in Analytical Biochemistry
- Vol. 449, 45-51
A chromogenic assay for limit dextrinase and pullulanase activity
- 1 March 2014
- Published by Elsevier in Analytical Biochemistry
- Vol. 449, 45-51
A chromogenic assay for limit dextrinase and pullulanase activity