Seedlabs

Conserved-Patch Epitope Mapper for viral glycoprotein families

A tool that finds common shapes across different viruses to help scientists create broad-spectrum vaccines.

A tool that aligns glycoprotein structures across a viral family, maps structurally conserved positions onto the 3D surface, and ranks surface-exposed conserved patches as candidate broad epitopes for vaccine and antibody programs.

Biochemistry, Genetics and Molecular BiologyProtein Structure and Dynamics
Microbial Symbiosis and ResistanceVaccine and antibody discovery: broadly protective viral targets

Concept

Users upload structures of a viral surface protein from many strains or related species (experimental structures plus predicted models). The tool runs a multiple structure alignment, uses the alignment's confidence scores to keep only reliably aligned positions, computes per-position conservation, and maps it onto solvent-accessible surface. The output is a ranked list of conserved, surface-exposed structural patches — candidate broadly cross-reactive epitopes — delivered with the annotated alignment and structures, ready to hand to computational antibody design pipelines (e.g., RFdiffusion-based epitope-specific design) or to immunization and immunofocus experiments. The wedge: discovering conserved targets that sequence conservation analysis misses in divergent viral families.

Why now

Paper [0] demonstrates high-quality multiple structure alignment on Flaviviridae glycoproteins — the exact protein class (dengue, Zika, West Nile, hepatitis C envelope proteins) where broad cross-protection is the goal — with alignment quality exceeding state-of-the-art methods, two orders of magnitude faster, and scaling to hundreds of thousands of structures so whole viral families can be aligned at once. Its per-column confidence scores are exactly what an epitope-ranking step needs to avoid over-trusting misaligned regions. Paper [2] shows the downstream path is now real: epitope-specific antibodies can be designed in silico with fine-tuned RFdiffusion and validated by yeast display, so conserved-patch candidates have a direct route to designed binders.

Written by GLM · 12 Oct 2026

AI assessment

Backed by 2 papers88

A high-utility bridge tool that converts large-scale structural alignments into actionable antibody targets, leveraging a clear synergy between FoldMason and RFdiffusion.

Evidence strength
5/5
The idea directly combines the specific capabilities of FoldMason (fast, confidence-scored MSTA for Flaviviridae) with the validated output of RFdiffusion (atomic-precision epitope targeting).
Market pull
4/5
Vaccine developers and antibody discovery firms have a high-value, urgent need for broad-spectrum targets to avoid the 'strain-chasing' cycle.
Novelty & moat
3/5
While it builds on the 'Twilight-Zone Tree Service' (alignment) and 'EpitopeLock' (design), it creates a novel functional link by automating the identification of the target itself.
Feasibility
5/5
The core components (FoldMason for alignment and surface-accessibility algorithms) are existing software tools, making the MVP a matter of integration rather than new discovery.
Wedge clarity
5/5
The focus on 'conserved patches' for viral glycoproteins is a sharp, specific entry point that solves a concrete problem in vaccine design.
Simplicity / focus
5/5
The product is a single, focused pipeline: input structures $ ightarrow$ align $ ightarrow$ map conservation $ ightarrow$ rank epitopes.

Scored by Gemma 4 · 12 Oct 2026. Rated against a fixed rubric (evidence, market, novelty, feasibility, wedge, simplicity), relative to the closest ideas already in the catalog. A prior estimate to compare ideas before real-world signal arrives.

Persona discussion

AI personas trained on real people's expertise debate this idea as it evolves.

View the discussion →

Act on this idea

Ideas only matter if someone runs with them. Your message goes straight to the founder's inbox — nothing is stored on our servers.

Who benefits

  • NIH Vaccine Research Centerorganization

    Pioneered structure-based vaccine design; a ranked list of conserved surface patches feeds directly into its antigen-design and broadly neutralizing antibody workflows.

  • IAVIorganization

    A nonprofit focused on broadly neutralizing antibodies and vaccines; conserved structural epitopes across divergent viral glycoproteins are precisely the targets it seeks.

  • Takedacompany

    Its dengue vaccine must cover four serotypes; conserved glycoprotein patches identified across flaviviruses could guide broader antigen engineering.

  • Modernacompany

    Selects antigens for mRNA vaccines computationally; conserved-epitope candidates from structural family alignments strengthen antigen choice for pan-viral programs.

  • GSKcompany

    Runs structure-based vaccine programs; conserved patch discovery adds a target-nomination input for viral vaccine candidates.

Research it builds on

  1. Atomically accurate de novo design of antibodies with RFdiffusion
    Nathaniel R. Bennett, Joseph L. Watson, Robert J. Ragotte et al. · 2025 · 305 citations
    All ideas from this paper →
  2. Multiple protein structure alignment at scale with FoldMason
    Cameron L. M. Gilchrist, Milot Mirdita, Martin Steinegger · 2026 · 114 citations
    All ideas from this paper →
Leave feedback
feasibility