ABodyBuilder: Automated antibody structure prediction with data–driven accuracy estimation

Abstract

Computational modeling of antibody structures plays a critical role in therapeutic antibody design. Several antibody modeling pipelines exist, but no freely available methods currently model nanobodies, provide estimates of expected model accuracy, or highlight potential issues with the antibody's experimental development. Here, we describe our automated antibody modeling pipeline, ABodyBuilder, designed to overcome these issues. The algorithm itself follows the standard 4 steps of template selection, orientation prediction, complementarity-determining region (CDR) loop modeling, and side chain prediction. ABodyBuilder then annotates the ‘confidence’ of the model as a probability that a component of the antibody (e.g., CDRL3 loop) will be modeled within a root–mean square deviation threshold. It also flags structural motifs on the model that are known to cause issues during in vitro development. ABodyBuilder was tested on 4 separate datasets, including the 11 antibodies from the Antibody Modeling Assessment–II competition. ABodyBuilder builds models that are of similar quality to other methodologies, with sub–Angstrom predictions for the ‘canonical’ CDR loops. Its ability to model nanobodies, and rapidly generate models (∼30 seconds per model) widens its potential usage. ABodyBuilder can also help users in decision–making for the development of novel antibodies because it provides model confidence and potential sequence liabilities. ABodyBuilder is freely available at http://opig.stats.ox.ac.uk/webapps/abodybuilder.

Modeling and docking of antibody structures with Rosetta

“…In addition to RosettaAntibody, there are three publicly accessible, fully automated web servers for antibody structure prediction: Kotai Antibody Builder 46,47 , Prediction of ImmunoGlobulin Structures (PIGS) 48 , and ABodyBuilder 49 . The performance of each method was discussed in the recent Antibody Modeling Assessment (AMA) 6 , except for ABodyBuilder, which was developed and benchmarked on the AMA antibodies ex post facto .…”

“…PIGS does not include backbone refinement, and as a result it returns structures very rapidly with minimal computational cost. ABodyBuilder includes extensive refinement and has recent developments 49 ; it is different in that it allows CDR templates of mismatched lengths 26 and exploits a full six-dimensional V L -V H determination strategy 7 . RosettaAntibody is unique in that its extensive conformational refinement focused in antibody degrees of freedom is designed and tested toward creating a structure that is at an energy minimum and appropriate for downstream applications including docking or design.…”

Oxetane Grafts Installed Site‐Selectively on Native Disulfides to Enhance Protein Stability and Activity In Vivo

“…Finally, analysis of secondary structural content by CD showed no significant differences between the original and stapled antibodies (Figure 3 e). MD simulations performed on a 3D model of DesAb‐Aβ 3‐9 , previously generated using ABodyBuilder,22 and on the stapled derivative DesAb‐Aβ 3‐9 ‐ 1 , suggest that, although the 3D structure is maintained upon the chemical modification, the oxetane motif provokes a slightly increase in the degree of flexibility (Supporting Information, Figure S31). Collectively, these data demonstrate the suitability of the oxetane motif to staple solvent accessible disulfide bonds on proteins with minimal secondary structure alterations.…”

Immunoglobulin gene analysis as a tool for investigating human immune responses

“…A recent paper that used the RosettaAntibody 3.0 antibody modeling protocol162 estimated that modeling of 2000 sequences took approximately 570 000 CPU hours163 so clearly there are challenges in the development of tools for structural calculations at a scale to match the available repertoire information. Of the large number of different tools currently available it seems that ABodyBuilder is the speediest, at 30 seconds per structure, which is around 567 CPU hours per thousand sequences 151, 164, 165…”