Published November 7, 2026 | Version v1.0

Which antibodies bind best to viral proteins? Validating an automated docking pipeline for antibody cross-reactivity across viral families

  • 1. Department of Bioinformatics and Genomics, University of North Carolina at Charlotte
  • 2. Department of Bioinformatics and Genomics and CIPHER Center, University of North Carolina at Charlotte

Description

Data for Oral Presentation: Validating an automated docking pipeline for antibody cross-reactivity across viral families.

Slides, figures, abstract, and references from the oral presentation given at SNCURCS 2026, the State of North Carolina Undergraduate Research and Creativity Symposium, held on the campus of the University of North Carolina at Charlotte on 7 November 2026.

Authors

Srikritha Kosuri. Undergraduate researcher. Department of Bioinformatics and Genomics, University of North Carolina at Charlotte, Charlotte, North Carolina, USA. ORCID: 0009-0007-4257-3797. Corresponding author, skosuri2@charlotte.edu.

Anastasiia Duchenko. Mentor. Department of Bioinformatics and Genomics, University of North Carolina at Charlotte, Charlotte, North Carolina, USA. ORCID: 0009-0000-4070-7666.

Denis Jacob Machado. Advisor. Department of Bioinformatics and Genomics and CIPHER Center, University of North Carolina at Charlotte, Charlotte, North Carolina, USA. ORCID: 0000-0001-9858-4515.

What this deposit contains

Kosuri_SNCURCS2026_slides.pdf is the presentation as delivered, 13 slides plus 4 appendix slides.

Kosuri_SNCURCS2026_abstract.pdf is the abstract as submitted, with the full conference details.

references.bib lists every work cited in the talk, in BibTeX format.

The five figures are provided as vector PDF files.

  • avoidrune_workflow.pdf is the AVOIDRUNE pipeline, from input structures through docking to the summary TSV.
  • research_approach.pdf is the study design, from the two input datasets through the cluster to the two readouts.
  • zika-ab-complex.pdf is the docked Zika virus envelope glycoprotein and antibody Fab complex, PDB 5JHL, with the predicted binding site and a close view of one hydrogen bond.
  • heatmap_subset.pdf is the predicted binding score for every antibody against every receptor, for six viral targets and the prion control.
  • boxplot_by_receptor_bw_subset.pdf is the predicted binding affinity by receptor target, with pairwise significance tests.

Summary of the work

AVOIDRUNE is a containerized pipeline for protein-protein docking at scale. It cleans and renumbers input structures, predicts active sites on both partners, docks with HADDOCK3, and estimates binding affinity with PRODIGY.

This project tested whether AVOIDRUNE holds up under load and whether its scores mean anything biologically. The benchmark dataset covered 12 viruses across Coronaviridae, Flaviviridae, and Orthomyxoviridae, four per family, with protozoan, bacterial, and prion structures as negative controls. A mock dataset repeated the run with missing files, malformed CSV, mismatched chain IDs, corrupt structures, case-mismatched filenames, empty files, and unusual character encodings.

Eight stress tests ran on the University of North Carolina at Charlotte HPC cluster through SLURM, using the Nebula and Tucana partitions on 48 AMD processors, and finished in 21 hours and 43 minutes using 8.0 GB of memory in total. Seven of 12 viral targets bound their matched antibody as expected, three strongly and four moderately. Negative controls bound weakly, and the prion protein was the most consistently weak binder. The pipeline stopped on missing key files and named what was absent, and it skipped only the broken files when the input was partly malformed.

A note on one figure. The color bar on heatmap_subset.pdf was relabelled to read the quantity alone. It originally read "PredGE (kcal/mol); <= -14 = strong", and that single cut point contradicts the three-tier scale used throughout the talk, where binding below -16 kcal/mol is strong, -16 to -14 is moderate, and -14 to -12 is weak. No data values were changed.

A note on the abstract. It is reproduced as submitted, so it is written in the future tense. The slides report what the completed runs showed.

Conference

SNCURCS 2026, the State of North Carolina Undergraduate Research and Creativity Symposium, in its 22nd year. Hosted by the University of North Carolina at Charlotte and co-sponsored with Central Piedmont Community College. Saturday, 7 November 2026, Popp-Martin Student Union, Mebane Hall, and the CHHS Building, 8845 Craver Road, Charlotte, North Carolina 28223, USA. Oral presentation of 10 to 12 minutes. Program at our.charlotte.edu/sncurcs.

How to cite

Kosuri S, Duchenko A, Jacob Machado D. 2026. Which antibodies bind best to viral proteins? Validating an automated docking pipeline for antibody cross-reactivity across viral families. Presented at SNCURCS 2026, Charlotte, North Carolina, 7 November 2026. Zenodo. DOI: 10.5281/zenodo.24088220.

License

Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0). Share and adapt this material for non-commercial purposes, credit the authors, and release anything you build from it under the same terms.

Contact

Srikritha Kosuri, skosuri2@charlotte.edu

Denis Jacob Machado, dmachado@charlotte.edu, https://phyloinformatics.com

Updates are linked at https://phyloinformatics.com/outreach/

Notes

Srikritha Kosuri gave this talk as an undergraduate researcher in the Phyloinformatics Lab, Department of Bioinformatics and Genomics, University of North Carolina at Charlotte. Anastasiia Duchenko mentored the project and Denis Jacob Machado supervised it. The analyses ran on the UNC Charlotte HPC cluster through University Research Computing, on the Nebula and Tucana partitions.

Files

Kosuri_SNCURCS2026_slides.pdf

Files (10.7 MB)

Name Size
md5:89e2e3e1f5dabd29a67dc40ae902105d
22.5 kB Preview Download
md5:312c2a0300d1ba876bb9cb8d8557bac7
46.1 kB Preview Download
md5:ed4a4a238f04c5d0570b0f255ef0b34b
700.1 kB Preview Download
md5:5548adcf018fce910ed126c97fe79f58
58.0 kB Preview Download
md5:eaf26af982761e2a6ee892e40ae2a52f
5.6 MB Preview Download
md5:8c1446d42cefbac5680ff997520dd28b
5.2 kB Download
md5:72f46e3789c519374fa37f15ca61bc39
31.8 kB Preview Download
md5:475eb43af0b3b0a1ea0e541c6935be66
4.3 MB Preview Download

Additional details

Related works

Is described by
Other: https://phyloinformatics.com/2026/11/07/KosuriSNCURCS.html (URL)
Is referenced by
Other: https://phyloinformatics.com/outreach/ (URL)
References
Journal article: 10.1016/j.chom.2016.04.013 (DOI)