Triple Vax

11 antigens · 4 constructs · local evidence portal

Overview Membrane Atlas Vaccine Builder Validation Dynamics Methods

Treponema denticola

Porphyromonas gingivalis

Tannerella forsythia

No antigens match this search.

Loading structure

Loading antigenPreparing the oriented structure and residue evidence.

Methods summary

What the portal computes, how evidence is gated, and what remains unproven

Computational triage only. The portal ranks candidates and exposes evidence; it does not establish immunogenicity, protection, safety, or clinical efficacy. Wet-lab validation is required.

Reading rule
A residue can be solvent-exposed on an isolated model without being antibody-reachable on the intact bacterial cell.

01 · Structure

Place each antigen in membrane context

AlphaFold-derived structures are oriented with a barrel-axis and hydrophobic-fit model. Shrake–Rupley accessibility (1.4 Å probe) and an asymmetric Gram-negative outer-membrane model classify exposed, pore-facing, lipid-facing, and periplasmic residues [1–5]. Side labels are withheld below 70% confidence.

02 · Antigen gates

Remove biologically inaccessible regions

Predicted Sec signal peptides are excluded from B-cell mining [6]. Localization, virulence, host non-homology, conservation, safety, and population coverage remain explicit evidence gates; missing evidence is reported as incomplete rather than assumed.

03 · Epitopes

Separate antibody and T-cell evidence

B-cell candidates combine hydrophilicity, flexibility, and RSA, then require the antibody-accessible surface [5,7]. MHC-I/II candidates use IEDB NetMHCpan percentile ranks and are not membrane-gated because they are processed intracellularly [8].

04 · Constructs

Assemble traceable multi-epitope vaccines

β-defensin, PADRE, HP91, and EAAAK/AAY/GPGPG linkers define the construct architecture [9–13]. Every epitope retains an exact link back to its source antigen and residue span.

05 · Validation

Check fold, chemistry, docking, and display

ProtParam-equivalent physicochemistry, fold confidence, TLR2 docking, and protruding-residue clustering provide orthogonal checks [14,15]. A folded construct is displayed only when its model sequence exactly matches the current construct sequence.

06 · Interpretation

Use confidence without overclaiming

High-confidence structure does not prove antigen accessibility or vaccine efficacy. The portal distinguishes standard calculations from transparent heuristics and keeps unavailable or stale evidence visibly unresolved.

Methods, metrics & references — the exact value behind every number

Each row is the constant actually used in the cited pipeline module, with the published method it comes from. References are numbered below.

Stage	Method	Metric & threshold (exact value used)	Ref
Orientation	PCA frame canonicalization → barrel-axis / hydrophobic-belt SVD fit (`orientor.py`, `barrel.py`)	Rotation-invariant: 5× random-rotation extracellular-set Jaccard ≥ 0.95 (all 11 antigens = 1.00). Barrel quality: ring-CV (roundness), lipid−pore ΔKD gate ≥ 0.5, aromatic Trp/Tyr girdle fraction.	3,5
Orientation	Surface-antigen anchor-relative outward axis (`orientor.py`)	Exposed at RSA ≥ 0.20; distal-axis fraction ≥ 0.5 = outward face.	2
Membrane	Asymmetric Gram-negative outer-membrane bilayer model (`membrane.py`)	Hydrophobic core 20–35 Å; interfacial band 8 Å/leaflet; LPS headgroup shield 4 Å; periplasmic buffer 1.5 Å.	1,4
Membrane	Kyte–Doolittle hydropathy + side-chain radial facing (`membrane.py`)	KD ∈ [−4.5, +4.5]; lipid-facing radial dot ≥ +0.20, pore/water-facing ≤ −0.20. Side (EC/periplasmic) labelled only when combined confidence ≥ 70%.	5
Accessibility	Shrake–Rupley SASA, 1.4 Å probe (`sasa.py`)	RSA = SASA ÷ Tien-2013 maximum allowed ASA; antibody-accessible gate RSA ≥ 0.20.	2,16
B-cell mining	Per-residue linear propensity (`mine_membrane_epitopes.py`)	P = 0.40·hydrophilicity + 0.20·flexibility + 0.40·RSA, each ∈ [0,1]; hydrophilicity = (4.5 − KD)/9, flexibility = Karplus–Schulz normalised.	5,7,2
B-cell mining	Sliding window + accessibility gate (`mine_membrane_epitopes.py`)	15-mer windows; β-barrel gate = antibody-accessible fraction ≥ 0.50 (LPS-shielded counts 0.5×); non-barrel surface gate = raw RSA ≥ 0.25; predicted Sec signal peptide excluded before mining.	6
Topology	Extracellular-side call (`topology.py`)	Loop-architecture: extracellular loop ≥ 8 residues, decisive median+long-loop margin ≥ 0.15; tiebreakers = positive-inside (Arg/Lys/His density) and N/C-terminus periplasmic prior.	6
T-cell	MHC-I / MHC-II prescreen	NetMHCpan-4.1 / NetMHCIIpan-4.0 percentile rank (lower = stronger); HLA-agnostic and not membrane-gated (peptides are processed intracellularly).	8
Conservation	Strain-ortholog conservancy (`analysis/conservation.py`)	Mean sequence identity % and conservancy % across strain orthologs; epitope retained on the mean-identity gate.	—
Construct	Adjuvant / helper / linker architecture (`construct/`)	β-defensin (TLR1/2 agonist) · HP91 (innate) · PADRE (universal CD4 helper) · EAAAK rigid / AAY / GPGPG / KK linkers; every epitope keeps an exact residue link to its parent antigen.	9,11,12,13
Validation	Physicochemistry + fold + innate docking + display (`validation_map.py`)	ProtParam-equivalent: instability index (stable < 40), GRAVY, aliphatic index, theoretical pI. Fold confidence: ESMFold/AlphaFold3 pLDDT & pTM. Innate docking: ClusPro TLR1/TLR2 (PDB 2Z81) center score + interface residues. Surface display: protruding-residue clustering.	14,15,17
Dynamics	Elastic-network flexibility — GNM / ANM (`dynamics/enm.py`)	Per-residue mean-square fluctuation (RMSF proxy) from inverse-eigenvalue-weighted modes; GNM cutoff 7.5 Å, ANM 15 Å; mode collectivity κ; GNM B-factor recovery (Pearson r) as a validation check. Structure-derived, not MD.	18,19,20
Dynamics	Receptor–vaccine interface (`dynamics/interface.py`)	Inter-body Cα contacts (8 Å), interface residues, buried SASA (Shrake–Rupley), and an ANM flexibility overlay (is the interface a rigid binding-competent scaffold).	16,19
Dynamics	TITO transfer-operator surrogate (`dynamics/timescales.py`)	Brownian-mode RELATIVE implied timescales (τ_i/τ_j = λ_j/λ_i, exact & calibration-free) + a parametric N-ns horizon. The absolute ns axis, macrostate transition matrix, and bound-state occupancy are NOT produced — they require MD + a trained operator (see the 200 ns MD→MSM scaffold).	21,19

Rigor ledger. Standard or first-principles components: SASA, membrane-fit objective, barrel classifier, outer-membrane zones, ProtParam, ClusPro interface, protrusion clustering, and IEDB MHC. Transparent heuristics: B-cell propensity, signal-peptide approximation, and the quarantined in-house Pg/Tf MHC scorer.

Full methodology references (21)

[1] Henderson JC, et al. (2016). The power of asymmetry: architecture and assembly of the Gram-negative outer membrane lipid bilayer. Annu Rev Microbiol 70:255-278.
[2] Tien MZ, et al. (2013). Maximum allowed solvent accessibilities of residues in proteins. PLoS ONE 8(11):e80635.
[3] Lomize AL, Todd SC, Pogozheva ID (2022). Spatial arrangement of proteins in planar and curved membranes by PPM 3.0. Protein Sci 31(1):209-220.
[4] Nikaido H (2003). Molecular basis of bacterial outer membrane permeability revisited. Microbiol Mol Biol Rev 67(4):593-656.
[5] Kyte J, Doolittle RF (1982). A simple method for displaying the hydropathic character of a protein. J Mol Biol 157(1):105-132.
[6] von Heijne G (1986). A new method for predicting signal sequence cleavage sites. Nucleic Acids Res 14(11):4683-4690.
[7] Karplus PA, Schulz GE (1985). Prediction of chain flexibility in proteins. Naturwissenschaften 72(4):212-213.
[8] Reynisson B, et al. (2020). NetMHCpan-4.1 and NetMHCIIpan-4.0. Nucleic Acids Res 48(W1):W449-W454.
[9] Biragyn A, et al. (2002). Toll-like receptor-dependent activation of dendritic cells by beta-defensin 2. Science 298(5596):1025-1029.
[10] Vartak A, Sucheck SJ (2016). Recent advances in subunit vaccine carriers. Vaccines 4(2):12.
[11] Livingston B, et al. (2002). A rational strategy to design multiepitope immunogens based on multiple Th lymphocyte epitopes. J Immunol 168(10):5087-5096.
[12] Alexander J, et al. (1994). Development of high potency universal DR-restricted helper epitopes by modification of high affinity DR-blocking peptides. Immunity 1(9):751-761.
[13] Arai R, et al. (2001). Design of the linkers which effectively separate domains of a bifunctional fusion protein. Protein Eng 14(8):529-532.
[14] Guruprasad K, et al. (1990). Correlation between stability of a protein and its dipeptide composition. Protein Eng 4(2):155-161.
[15] Ikai A (1980). Thermostability and aliphatic index of globular proteins. J Biochem 88(6):1895-1898.
[16] Shrake A, Rupley JA (1973). Environment and exposure to solvent of protein atoms. Lysozyme and insulin. J Mol Biol 79(2):351-371.
[17] Kozakov D, et al. (2017). The ClusPro web server for protein-protein docking. Nat Protoc 12(2):255-278.
[18] Bahar I, Atilgan AR, Erman B (1997). Direct evaluation of thermal fluctuations in proteins using a single-parameter harmonic potential. Fold Des 2(3):173-181.
[19] Atilgan AR, et al. (2001). Anisotropy of fluctuation dynamics of proteins with an elastic network model. Biophys J 80(1):505-515.
[20] Brüschweiler R (1995). Collective protein dynamics and nuclear spin relaxation. J Chem Phys 102:3396-3403.
[21] Hinsen K, et al. (2000). Harmonicity in slow protein dynamics. Chem Phys 261:25-37.

Targeting the Periodontal “Red Complex”

Eleven outer-membrane & surface antigens across the three keystone Gram-negative anaerobes of chronic periodontitis — each oriented in its asymmetric outer membrane, then mined for antibody-accessible epitopes.

Treponema denticola

Spirochete · motile · endoflagella

Porphyromonas gingivalis

Gram– coccobacillus · black-pigmented · keystone

Gingipain R1 OstA PorV SusC/RagA

Tannerella forsythia

Gram– fusiform · S-layered

BspA TfsA TfsB TonB receptor

Enter the Membrane Atlas →or click any antigen above to open it directly

How the platform works

Outer-membrane architecture — where epitopes come from

Now 6 / 11 antigens are true transmembrane β-barrels; the other 5 are surface, S-layer or secreted. Both expose loops to circulating antibody.

TM β-barrel OMP (6) Surface / S-layer / secreted (5)

Vaccine construct architecture

One chimeric multi-epitope antigen — adjuvant-flanked epitope blocks joined by rigid (EAAAK) and flexible (GPGPG / AAY / GGGS) linkers. Click the map to assemble it in 3D.

β-defensinTLR1/2 · N-term

B-cell epitopes·GPGPG·

MHC-II·GPGPG·

PADRECD4 helper

N-TERMINUSC-TERMINUS

Each organism gets its own arm (Pg · Tf); the combined RedVax-Chimera stacks all three (Td + Pg + Tf) into one ESMFold-foldable construct. B-cell picks are membrane-gated; MHC-I/II from IEDB NetMHCpan percentile rank.

Build it in the Vaccine Builder →

Socransky “red complex.” All antigens are computational candidates pending wet-lab validation. Localization split per docs/MEMBRANE.md; epitope selection per the td-pipeline / pg-tf-panel / chimera stages.