Disjoint combinations profiling (DCP): a new method for the prediction of antibody CDR conformation from sequence.
- Published
- Accepted
- Subject Areas
- Bioinformatics, Computational Biology, Molecular Biology, Immunology
- Keywords
- conformational prediction, CDR conformation, blind test, canonical templates, CDR-H3 sequence rules, DCP, humanisation, prediction from sequence, antibody engineering
- Copyright
- © 2014 Nikoloudis et al.
- Licence
- This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, reproduction and adaptation in any medium and for any purpose provided that it is properly attributed. For attribution, the original author(s), title, publication source (PeerJ PrePrints) and either DOI or URL of the article must be cited.
- Cite this article
- 2014. Disjoint combinations profiling (DCP): a new method for the prediction of antibody CDR conformation from sequence. PeerJ PrePrints 2:e292v2 https://doi.org/10.7287/peerj.preprints.292v2
Abstract
The accurate prediction of the conformation of Complementarity-Determining Regions (CDRs) is important in modelling antibodies for protein engineering applications. Specifically, the Canonical paradigm has proved successful in predicting the CDR conformation in antibody variable regions. It relies on canonical templates which detail allowed residues at key positions in the variable region framework or in the CDR itself for 5 of the 6 CDRs. While no templates have as yet been defined for the hypervariable CDR-H3, instead, reliable sequence rules have been devised for predicting the base of the CDR-H3 loop. Here a new method termed Disjoint Combinations Profiling (DCP) is presented, which contributes a considerable advance in the prediction of CDR conformations. This novel method is explained and compared with canonical templates and sequence rules in a 3-way blind prediction. DCP achieved 93% accuracy over 951 blind predictions and showed an improvement in cumulative accuracy compared to predictions with canonical templates or sequence-rules. In addition to its overall improvement in prediction accuracy, it is suggested that DCP is open to better implementations in the future and that it can improve as more antibody structures are deposited in the databank. In contrast, it is argued that canonical templates and sequence rules may have reached their peak.
Author Comment
A paper presenting a new classification of antibody CDR conformations, upon which the present methods' training/updating was based, is conjointly submitted to PeerJ preprint server, titled:"A complete, multi-level conformational clustering of antibody complementarity-determining regions." Version 2: added DOI of conjoint paper.
Supplemental Information
Appendix with collection of tables outlining the detected multi-conformation full-rogue clusters.
Notable features include resolutions close to 3Å and R-free > 0.25.
Appendix: absolute probabilistic significance of IF fragment disjointness.
Presentation of a probabilistic closed-form equation for selecting only statistically significant signature signals.
Detailed canonical templates by CDR/Length.
Canonical templates were derived from the clustering set for every applicable conformational cluster, using the definitions of structurally-determining residues described in Martin & Thornton, (1996).
Individual predictions per CDR.
Detailed tables with all predictions for every CDR in the test sets, along with a measure of RMSD distance of the Query conformation from the closest cluster medoid.