The Antibody Society

the official website of the antibody society

An international non-profit supporting antibody-related research and development.

You are here: Home / Archives for Bioinformatics

Application of Machine Learning and Informatics in Antibody and Protein Research

by

Antibody Engineering & Therapeutics, held in December 2021, offered many opportunities to hear exciting and informative presentations by experts in the field. We are pleased to present here a summary of a plenary lecture by Prof. Charlotte Deane (University of Oxford), kindly written by Dr. Czeslaw Radziejewski.

Application of Machine Learning and Informatics in Antibody and Protein Research
Charlotte Deane, Professor of Structural Bioinformatics, Department of Statistics, University of Oxford

Machine learning relies heavily on the availability of large databases. Three databases for antibody research were developed in Prof. Dean’s lab: OAS (Observed Antibody Space),[1] SAbDab (Structural Antibody Database),[2] and Thera-SAbDab (database of immunotherapeutic variable domain sequences). OAS contains about 2 billion redundant antibody sequences across diverse immune states, organisms, and individuals. SAbDab is a fully automated self-updating collection of publicly available antibody structure data. It contains 5650 structures, but about 1000 truly non-redundant structures, 4213 antigen-antibody complexes and 890 structures of nanobodies. Thera-SabDob contains 696 structures as of October 2021. In addition, the lab has a CoV-AbDab database that contains sequences and structures for coronavirus antibodies for SARS-CoV-2, SARS-CoV-1 and MERS-CoV. This database contains about 5000 data points. The lab developed the SAbPred suite of tools for antibody prediction, comprising AntibodyBuilder, SPHINX, SCALOP, PEARS, ANARCI, ABangle, Hu-mAb SAAB+,TAP, Epitope Profiling SPACE and Ab-Ligaty. SCALOP, ABodybuilder, SPHINX are designed for building antibody models. ABlooper tool builds complementary-determining region (CDR) structures. ABangle is a tool for calculating and analyzing the VH-VL orientation in antibodies. TAP (Therapeutic Antibody Profiler) considers the drug-like properties of therapeutic antibodies.[3] It evaluates variable domains in antibody of interest using five developability criteria derived from post clinical Phase 1 antibody therapeutics. Epitope Profiling-SPACE and Paratyping Ab-Ligity can used to determine if two antibodies with divergent sequences can bind to the same epitope.[4] ANARCI is a tool for annotating antibody sequences and Hu-Mab is a computational tool for antibody humanization. Dlab is a deep learning method for virtual screening of antibody sequences that can bind specific antigens.

Professor Deane provided examples of using some of her computational tools. Antibody humanization is currently inefficient, as it is carried out experimentally in a largely trial and error process. Applying machine learning to an edited OAS database (with redundancies removed) led to classifiers that could distinguish between human and non-human antibody variable domain sequences. These classifiers were used to create the computational humanization tool Hu-mAb. Available sequences of therapeutic antibodies from different stages of development were subjected to Hu-mAb analysis. The high Hu-mAb scores correlated with low observed immunogenicity of an antibody and low scores correlated with higher observed immunogenicity. Twenty-five experimentally humanized antibody sequences for which rodent or rabbit precursor sequences were available were assessed by Hu-mAb. Most of the mutations that Hu-mAb generated were either the same or chemically similar for VH (77% and 85%, respectively) and for VL (59% and 58%, respectively). Hu-mAb suggested overall fewer mutations and fewer mutations to VH-VL interface than the experimental approach, therefore such humanized antibodies would more likely have preserved structure and function.

The Therapeutic Antibody Profiler evaluates properties thought to determine antibody developability, including CDRH3 or total CDR length; patches of surface hydrophobicity across CDR vicinity; patches of positive charges and negative charges across CDR vicinity; and structural Fv charge symmetry. These properties are related to aggregation, viscosity, poor expression and polyspecificity of antibody molecules.[5] TAP was applied in a study that used 137 post Phase 1 therapeutic models,14000 representative Human Antibody Models and 2 datasets of MedImmune Developability Failures. The study revealed that therapeutic antibodies tend to have shorter CDRH3 and smaller hydrophobic patches than natural ones. However, positive and negative patches of natural and therapeutic antibodies have similar profiles and Fv charge symmetry is also very similar. Both therapeutic and natural antibodies have an aversion to strongly oppositely charged VH and VL chains.

ABlooper [6] uses similar architecture as AlphaFold. It predicts structures of all six CDR loops and estimates the accuracy of prediction. The root-mean-square deviation from AlphaFold2 for CDRH3 prediction (2.87A) were comparable with ABlooper (2.49 A). Unlike AlphaFold2, ABlooper generates a series of predicted structures from which a prediction of accuracy can be estimated. If the predicted structures are widely divergent, then the quality of prediction is low. ABlooper is also much faster than other deep learning methods such as AlphaFold (100 structures predicted in 5 second vs one structure in 20 min). All tools are available freely for academic institutions.

  1. Olsen TH, Boyles F, Deane CM. Observed Antibody Space: A diverse database of cleaned, annotated, and translated unpaired and paired antibody sequences. Protein Sci. 2022 Jan;31(1):141-146. doi: 10.1002/pro.4205.
  2. Schneider C, Raybould MIJ, Deane CM. SAbDab in the age of biotherapeutics: updates including SAbDab-nano, the nanobody structure tracker. Nucleic Acids Res. 2022 Jan 7;50(D1):D1368-D1372. doi: 10.1093/nar/gkab1050.
  3. Raybould MIJ, Deane CM. The Therapeutic Antibody Profiler for Computational Developability Assessment. Methods Mol Biol. 2022;2313:115-125. doi: 10.1007/978-1-0716-1450-1_5.
  4. Wong et al. Ab-Ligity: identifying sequence-dissimilar antibodies that bind to the same epitope. MAbs 2021. DOI: 10.1080/19420862.2021.1873478.
  5. Khetan et al. Current advances in biopharmaceutical informatics: guidelines, impact and challenges in the computational developability assessment of antibody therapeutics. MAbs 2022. DOI: 10.1080/19420862.2021.2020082.
  6. Abanades B, Georges G, Bujotzek A, Deane CM. ABlooper: Fast accurate antibody CDR loop structure prediction with accuracy estimation. Bioinformatics. 2022 Jan 31:btac016. doi: 10.1093/bioinformatics/btac016.

Learn about Modeling Biologic Molecules on January 27th!

by

Registration is open for our next webinar, “Modeling Biologic Molecules“, to be held Thursday January 27, 2022, 11am ET / 5pm CET.

The speakers are Dr. Monica Fernández-Quintero and Prof. Klaus R. Liedl.

Modeling in Chemistry obviously depends on a strong link to reality. Even though the mathematical description of chemistry has been possible for almost 100 years, realistic modelling has only recently become available due to the recent massive increase of computing power following Moore’s law. Still, appropriate statistics, initial conditions and boundaries pose considerable challenges. Nowadays, methodological advances and progress in hardware allows the observation of biological systems for relevant time periods. Hence, dynamic processes like reorientations, folding and binding can be seen in atomistic resolution leading to completely new insights.

Describing an antibody’s binding site using only one single static structure limits the understanding and characterization of the antibody’s function and properties, whereas various biophysical properties are governed by its dynamics, e.g., antibody-antigen binding. This limitation is even more pronounced when no experimentally determined structure is available or the crystal structure is distorted by packing effects, which can result in misleading antibody paratope structures. To improve antibody structure prediction and to take the strongly correlated CDR loop and interface movements into account, antibody paratopes should be described as interconverting states in solution with varying probabilities. These kinetically characterized paratope ensembles with their respective state probabilities allow the identification of the dominant conformation in solution, which frequently has been shown to coincide with the binding competent conformation. Therefore, the definition of kinetically and functionally relevant states, so-called paratope states, can be successfully used to improve the accuracy and enhance the predictivity of antibody-antigen docking.

Register for this webinar here!

AIRR Community Standards v1.3 and COVID-19 AIRR-Seq Data Available

by

The AIRR Community, a grassroots group of immunologists, immunogeneticists and computer scientists who are dedicated to developing protocols and standards to facilitate sharing of these data is pleased to announce the release of v1.3 of the AIRR Standards, including v1.0 of the AIRR Data Commons API (ADC API) to query AIRR compliant data repositories. The AIRR Community also notes that the first repositories in the AIRR Data Commons, based completely on AIRR Standards, are now in production, with the iReceptor Public Archive and VDJServer repositories early adopters of these AIRR Standards. With the release of v3.0 of the iReceptor Scientific Gateway, it is possible for researchers to query the AIRR Data Commons, greatly expanding the ability of researchers to find and analyze AIRR-seq data in support of biomedical research and improving patient care.

The AIRR Community is also pleased to announce the availability of the first COVID-19 AIRR-seq data sets in the AIRR Data Commons. The COVID-19 data is from the Nielsen et al. paper and is currently stored in the community based AIRR COVID-19 repository and is searchable through the iReceptor Scientific Gateway (gateway.ireceptor.org). More studies from COVID-19 patients will be added in the coming weeks. The AIRR Community is tracking and curating COVID-19 AIRR-seq papers and data sets, including their availability in the AIRR Data Commons, on the b-t.cr web site (https://b-t.cr/t/publicly-available-covid-19-airr-seq-data-sets).

Finally, the Research Data Alliance recently released its “guidelines and recommendations on data sharing in the context of COVID-19“. AIRR Community members Brian Corrie and Christian Busse participated in the preparation of this document, and in particular the authoring of the section around recommendations for sharing COVID-19 AIRR-seq data.

COVID-19 Demands Increased Public Sharing of Biomedical Research Data

by

Defeating the coronavirus pandemic will require unprecedented cooperation from the research community. This is especially true for the Adaptive Immune Receptor Repertoire Community (AIRR-C), given the paramount importance of antibodies and T cells for vaccines, diagnostics, and therapeutics in viral infection. Therefore, the AIRR-C hereby calls upon its members, and the wider research community, to share experiences, resources, samples, and data as openly and freely as possible, and to work within their respective systems to break down barriers to achieve this goal, subject to the overarching directives of respect, privacy, and protection for patients and all people. We are in this together.

Cyrus Chothia – In Memoriam

by

Post contributed by: Prof. Andrew C.R. Martin, Professor of Bioinformatics and Computational Biology; SMB Graduate Tutor
Institute of Structural and Molecular Biology, University College London

Cyrus Chothia, FRS, (19 February 1942 – 26 November 2019), was an emeritus scientist at the Medical Research Council Laboratory of Molecular Biology (LMB) in Cambridge, UK where he was a fellow of Wolfson College. He studied at Durham University, followed by an MSc at Birkbeck College, University of London, a small college but famed for its involvement in the development of structural biology and x-ray crystallography, being the home of such luminaries as J.D. Bernal, Aaron Klug and Rosalind Franklin. This was followed by a PhD at University College London supervised by Peter Pauling, son of Linus Pauling. Cyrus was one of the founding fathers of structural bioinformatics and made a particular contribution in the antibody field. Amongst others, he worked with Nobel prize winner, Michael Levitt, Joel Janin, Alexey Murzin, Tim Hubbard and Anna Tramontano, but he is perhaps best known for his work with Arthur Lesk. His PhD students included a number of people who have gone on to make major contributions in bioinformatics, such as Alex Bateman, Steve Brenner, Mark Gerstein, Julian Gough, Sarah Teichmann, and Bissan Al-Lazikani.

Back in the early 1970s, Wu and Kabat had demonstrated the presence of hypervariable sequence regions in antibody variable domains that they suggested would form structural loops or complementarity-determining regions (CDRs) that come together in 3D to form the binding site. This was confirmed when Poljak solved the first antibody crystal structure, and it was assumed that the CDRs would also be extremely variable in structure. I first met Cyrus in the late 1980s when I was doing my DPhil in Oxford on modelling antibody combining sites with Anthony Rees. By that time, around eight structures of antibodies, or Bence Jones light chain dimers, were available and Cyrus, together with Arthur Lesk, compared these. They found that, with the exception of the third CDR of the heavy chain (CDR-H3), the structures of the remaining CDRs were remarkably conserved. Further they proposed that the presence of certain ‘key residues’ – either within the CDRs, or packing against them – would define the conformation [PMID: 3090684, 3681981, 2687698]. To be frank, Tony and I didn’t really believe it. After all, there were potentially billions of antibody sequences and they had looked at fewer than 10. Cyrus and Arthur came to visit us in Oxford, and I remember sitting in The Eagle and Child with them discussing these ideas. Cyrus was always modest and completely accepted that they may be wrong, but of course they turned out to be largely correct. As more structures became available, the rules evolved with the importance of other positions being recognized [PMID: 2118959], but the principle was completely upheld. When we published a paper on key residues in 1986, I spoke to him about one of the outliers that appeared not to follow the rules. His view (which was almost certainly correct!) was that the crystal structure was wrong. More recent analysis by ourselves and others has suggested that the rules aren’t always as precise as might once have been thought and the requirements for framework mutations outside the key residues in order to achieve good binding in antibody humanization supports the view that the precise conformation is influenced by other residues and the detail of the environment around the CDRs.

Cyrus introduced a definition of the ‘structural loops’ in antibodies. These are frequently referred to as the ‘Chothia CDRs’, a term that he did not like as, in his view, the CDRs were the sequence-variable regions defined by Wu and Kabat, while his definition related to regions that were structurally variable and he would not presume to redefine what had been done by Wu and Kabat. In fact, his definitions changed over his various papers as more structural information became available. He also introduced the Chothia numbering scheme for antibodies, which was based on Kabat numbering but corrected the insertion sites in CDR-L1 and CDR-H1 to be structurally correct. Unfortunately in 1989, they made an error such that the insertions in CDR-L1 were placed after residue L31 rather than L30. As another example of his humility and modesty, I happened to referee a paper of theirs in 1997 and recognized this error. He guessed that I was the referee, contacted me, and immediately accepted the correction.

While I have focussed on his work on antibodies, he was widely known for his work in many areas of understanding protein structure. He was elected as a Fellow of the Royal Society (FRS) in 2000, for having “shown how the amino sequences of proteins determine their structure, function and evolution”. To name just a few of his contributions, he developed the SCOP classification of protein structure with Alexey Murzin [PMID: 7723011] and the SUPERFAMILY database with Julian Gough. He studied multi-domain proteins [PMID: 15093836], protein packing [PMID: 10388571] and was involved in functional annotation of more than 60,000 cDNAs from the mouse transcriptome [PMID: 12466851]. As well as his work on the conformation of the CDRs, he examined the packing of VH and VL domains in antibodies [PMID: 4093982] and examined the evolution of immunoglobulin domains in general [PMID: 7175935]. With Bissan Al-Lazikani, he published his final paper on canonical classes of antibody CDRs [PMID: 9367782], but then extended this into T-cell alpha-beta receptors [PMID: 10656805].

Cyrus made enormous contributions to our understanding of protein evolution in general as well as of the structure of antibodies. He will be hugely missed by the scientific community, and by me personally. His science and the many well-known scientists who did their PhDs with him or were influenced by him, are a huge and lasting legacy.