Design of full-length agonist and antagonist antibodies for hard targets
- 1Pythia Labs
Abstract
Therapeutic antibodies have transformed modern medicine, yet conventional discovery relies on lengthy experimental workflows, offers limited control over how antibodies engage their targets, and struggles to deliver candidates across many important target classes. Despite recent advances in computational protein design, generating full-length antibodies with desired functional activity and favorable developability profiles remains challenging. Here we present an end-to-end computational design pipeline built around Pythia-1, an all-atom generative model that reasons jointly over amino acid sequence and three-dimensional atomic structure under explicit design constraints. We generated candidates in full-length human immunoglobulin G (IgG) format across 22 therapeutic targets spanning receptor ectodomains, cytokines, growth factors, and G protein-coupled receptors (GPCRs), a particularly challenging target class for antibody discovery. Following a single computational design round, experimental screening of at most 48 candidates per target identified antibodies with single-digit nanomolar affinities for soluble proteins and single-digit nanomolar cell-binding EC50 values for GPCRs. The pipeline's ability to target specified epitopes and conformational states enabled the design of antibodies with agonist and antagonist activity. In vitro characterization demonstrated target-specific engagement and potent functional modulation, with single-digit nanomolar inhibitory potency in functional assays for a range of targets, including multiple GPCRs. To our knowledge, this study provides the first demonstration of both GPCR agonism and antagonism by de novo designed full-length IgG antibodies. We incorporated key developability and manufacturability requirements into the initial designs by conditioning generation on human germline frameworks and applying sequence constraints to limit known liabilities. The resulting antibody designs exhibited biophysical profiles comparable to or better than those of clinical-stage therapeutic controls. We provide the complete sequences of all lead candidates to support independent experimental validation and further development. Together, these findings demonstrate that our technology can deliver antibodies with programmable function and key properties expected of lead candidates, opening opportunities for first-in-class therapeutics against hard-to-drug targets.
Related
- High-throughput affinity profiling of antibodies against soluble and native membrane targets by digital flow cytometry
- A genome-wide computational approach for therapeutic targets identification and synthetic peptide assembly enable rational vaccine design against Arcanobacterium haemolyticum
- De novo design of monoclonal and bispecific antibodies with OFAntibody
- ProMORNA: An End-to-End Generative Framework for Full-Length mRNA Design
- Combined SGLT-2 Inhibitor and Mineralocorticoid Receptor Antagonist Therapy: Effects on the Renin–Angiotensin–Aldosterone System in Heart Failure Treatment
BibTeX
@misc{team2026design,
title = {{Design of full-length agonist and antagonist antibodies for hard targets}},
author = {Pythia Labs Team and R. A. Aglietti and R. Ansari and P. M. Bowers and T.-H. Buelles and H. Chauvin and T. M. Duplay and M. El Hibouri and J. Jorda and J. Laniado and M. M. A. Malago and K. Meador and A. Polovina and K. Polyak and J. Sosna and R. Tacke and T. O. Yeates and L. Zanini},
howpublished = {bioRxiv},
year = {2026},
month = oct,
doi = {10.64898/2026.10.21.760573},
url = {https://www.biorxiv.org/content/10.64898/2026.10.21.760573v1},
note = {CC BY-NC},
}
RIS
TY - UNPB
TI - Design of full-length agonist and antagonist antibodies for hard targets
AU - Pythia Labs Team
AU - R. A. Aglietti
AU - R. Ansari
AU - P. M. Bowers
AU - T.-H. Buelles
AU - H. Chauvin
AU - T. M. Duplay
AU - M. El Hibouri
AU - J. Jorda
AU - J. Laniado
AU - M. M. A. Malago
AU - K. Meador
AU - A. Polovina
AU - K. Polyak
AU - J. Sosna
AU - R. Tacke
AU - T. O. Yeates
AU - L. Zanini
PY - 2026
DA - 2026/10/21
PB - bioRxiv
DO - 10.64898/2026.10.21.760573
UR - https://www.biorxiv.org/content/10.64898/2026.10.21.760573v1
AB - Therapeutic antibodies have transformed modern medicine, yet conventional discovery relies on lengthy experimental workflows, offers limited control over how antibodies engage their targets, and struggles to deliver candidates across many important target classes. Despite recent advances in computational protein design, generating full-length antibodies with desired functional activity and favorable developability profiles remains challenging. Here we present an end-to-end computational design pipeline built around Pythia-1, an all-atom generative model that reasons jointly over amino acid sequence and three-dimensional atomic structure under explicit design constraints. We generated candidates in full-length human immunoglobulin G (IgG) format across 22 therapeutic targets spanning receptor ectodomains, cytokines, growth factors, and G protein-coupled receptors (GPCRs), a particularly challenging target class for antibody discovery. Following a single computational design round, experimental screening of at most 48 candidates per target identified antibodies with single-digit nanomolar affinities for soluble proteins and single-digit nanomolar cell-binding EC50 values for GPCRs. The pipeline's ability to target specified epitopes and conformational states enabled the design of antibodies with agonist and antagonist activity. In vitro characterization demonstrated target-specific engagement and potent functional modulation, with single-digit nanomolar inhibitory potency in functional assays for a range of targets, including multiple GPCRs. To our knowledge, this study provides the first demonstration of both GPCR agonism and antagonism by de novo designed full-length IgG antibodies. We incorporated key developability and manufacturability requirements into the initial designs by conditioning generation on human germline frameworks and applying sequence constraints to limit known liabilities. The resulting antibody designs exhibited biophysical profiles comparable to or better than those of clinical-stage therapeutic controls. We provide the complete sequences of all lead candidates to support independent experimental validation and further development. Together, these findings demonstrate that our technology can deliver antibodies with programmable function and key properties expected of lead candidates, opening opportunities for first-in-class therapeutics against hard-to-drug targets.
ER -