MIMOSA: Guiding De Novo Antibody Design with Natural Interaction Fingerprints

Abstract

De novo antibody design promises to transform therapeutic antibody and nanobody discovery. Yet, success remains inconsistent across targets, with generative models often yielding no experimentally validated binders. Here we introduce MIMOSA (MIMic-Oriented Structural Antibody generation), a model-agnostic, inference-time framework that constrains diffusion-based antibody design pipelines to reproduce the interaction geometry and chemistry of a target's known cognate binder interface. Rather than searching for a productive interface from scratch, MIMOSA constrains generation around residue identities and geometries already known to support binding, while allowing the generative model to complete the surrounding antibody interface. We test MIMOSA on different interface topologies: contiguous motifs that fit within a single CDR loop and spatially dispersed hotspots distributed across the paratope. Applied without retraining to two architecturally distinct generators (RFAntibody and BoltzGen), this framework delivers experimentally validated binders-achieving sub-micromolar to single-digit nanomolar affinities-at 20-26% hit rates on three targets on which unconstrained methods produce zero or near-zero binders: KEAP1, uPA, and IL-8. By turning any target with a structurally characterised cognate binder into an accessible de novo design problem, this framework broadens the practical reach of generative antibody design to targets where unconstrained approaches currently fail.

Auditing Protein-Protein Interaction Signals with Sparse Autoencoder Fingerprints
W. Zhu, S. Wang, X. Liu +4
bioRxiv27 Sep 2026PreprintCC BYFull textBioinformatics & computational biologybioRxiv
Interaction with complement factor C1q can contribute to and improve monoclonal antibody-mediated protection against lethal HSV-1 challenge
L. M. Jimenez, M. D. Slein, D. A. Leib +1
bioRxiv19 Oct 2026PreprintCC BY-NCFull textImmunologybioRxiv
MIMoSA: A Tool for Model-Independent Comparison of Transcription Factor Binding Motifs
A. V. Tsukanov, V. G. Levitsky
bioRxiv13 May 2026PreprintCC BYFull textBioinformatics & computational biologybioRxiv
Programmable Biochemical Logic Computation Through Repurposing Natural bZIP Protein Interaction Networks
A. Orozco-Estrada, D. Flores-Nuno, G. Mendizabal-Ruiz +2
bioRxiv29 Jul 2026PreprintCC BY-NCFull textSystems & synthetic biologybioRxiv
Brazilian brown propolis from Mimosa tenuiflora prevents airway smooth muscle hyper-responsiveness in an ovalbumin-induced asthma model
Isabela Motta Felício, Alissa Maria de Oliveira Martins, Alexya Mikelle Teixeira Cavalcanti +8
Phytochemistry Letters25 Sep 2026Peer-reviewedCC BY 4.0Full textZoology & animal behaviourHAL

Cite

BibTeX RIS

BibTeX

@misc{abanades2026mimosa,
  title = {{MIMOSA: Guiding De Novo Antibody Design with Natural Interaction Fingerprints}},
  author = {B. Abanades and A. Roncoli and M. Bhagawati and P. Kessel and S. Imhof-Jung and D. Doerr and J. Schilz and S. Vasilaki and F. Seeger and A. M. J. J. Bonvin and R. Bonneau and V. Gligorijevic and A. Vangone},
  howpublished = {bioRxiv},
  year = {2026},
  month = sep,
  doi = {10.64898/2026.09.24.753155},
  url = {https://www.biorxiv.org/content/10.64898/2026.09.24.753155v1},
  note = {CC BY},
}

RIS

TY  - UNPB
TI  - MIMOSA: Guiding De Novo Antibody Design with Natural Interaction Fingerprints
AU  - B. Abanades
AU  - A. Roncoli
AU  - M. Bhagawati
AU  - P. Kessel
AU  - S. Imhof-Jung
AU  - D. Doerr
AU  - J. Schilz
AU  - S. Vasilaki
AU  - F. Seeger
AU  - A. M. J. J. Bonvin
AU  - R. Bonneau
AU  - V. Gligorijevic
AU  - A. Vangone
PY  - 2026
DA  - 2026/09/24
PB  - bioRxiv
DO  - 10.64898/2026.09.24.753155
UR  - https://www.biorxiv.org/content/10.64898/2026.09.24.753155v1
AB  - De novo antibody design promises to transform therapeutic antibody and nanobody discovery. Yet, success remains inconsistent across targets, with generative models often yielding no experimentally validated binders. Here we introduce MIMOSA (MIMic-Oriented Structural Antibody generation), a model-agnostic, inference-time framework that constrains diffusion-based antibody design pipelines to reproduce the interaction geometry and chemistry of a target's known cognate binder interface. Rather than searching for a productive interface from scratch, MIMOSA constrains generation around residue identities and geometries already known to support binding, while allowing the generative model to complete the surrounding antibody interface. We test MIMOSA on different interface topologies: contiguous motifs that fit within a single CDR loop and spatially dispersed hotspots distributed across the paratope. Applied without retraining to two architecturally distinct generators (RFAntibody and BoltzGen), this framework delivers experimentally validated binders-achieving sub-micromolar to single-digit nanomolar affinities-at 20-26% hit rates on three targets on which unconstrained methods produce zero or near-zero binders: KEAP1, uPA, and IL-8. By turning any target with a structurally characterised cognate binder into an accessible de novo design problem, this framework broadens the practical reach of generative antibody design to targets where unconstrained approaches currently fail.
ER  -

Abstract

Related

Cite