Interactive 3D structural viewer for the human DNA Polymerase epsilon (Pol e) catalytic subunit (p261, UniProt Q07864).
This tool renders a procedural cartoon-ribbon schematic of the POLE holoenzyme directly in the browser using Three.js (WebGL). It is designed for exploring the domain architecture, catalytic mechanism, cancer-associated mutations, and clinical consequences of rare germline variants of Pol e.
Important: This is a schematic teaching model, not an atomic-coordinate viewer. The ribbon geometry, domain centers, and secondary-structure elements are procedurally generated to illustrate domain topology and functional relationships. Color modes (B-factor, conservation, charge, pathogenicity) display approximate domain-level or synthetic per-residue values, not crystallographic or computed data. For analysis of real atomic coordinates, load PDB 9F6D (Roske & Yeeles 2024, human Pol e-PCNA-DNA) into Mol* or PyMOL.
- Cartoon ribbon representation - helices, beta-strands (with arrowheads), and loops rendered per secondary structure element across all eight functional domains (NTD, EXO, Palm, P, Fingers, Thumb, Inactive-Pol, CTD)
- Two-metal-ion catalysis - ball-and-stick active site with octahedral Mg2+ coordination, conserved aspartate residues (D640, D642 [Motif A], D860 [Motif C]; human POLE1 numbering, UniProt Q07864 — verify against 9F6D coordinates before use; see note below), and distance annotations
- Incoming dNTP + Watson-Crick base pairing - template nucleotide with hydrogen bond visualization
- 3'-to-5' exonuclease site - two-metal proofreading active site with catalytic residues D275, E277 (ExoI, metal-coordinating, verified); D368 (ExoII, structural/DNA-binding — not a metal-coordinating catalytic residue); D462 (ExoIII, metal-coordinating, verified from PDB 9F6L mismatch excision structure, Roske & Yeeles 2024)
- Zinc-finger motifs - CysA/CysB in the CTD. CysA binds Zn2+ (required for PCNA binding). CysB is rendered as Zn2+ per the human p59-p261 crystal structure, though Netz et al. (2012) originally assigned a [4Fe-4S] cluster to CysB in B-family polymerases — the viewer depicts the Zn model; see the [4Fe-4S] entry for the CysX/P-domain cluster
- [4Fe-4S] cluster - iron-sulfur cluster in the P domain via the CysX motif (coordinating cysteines C651, C654, C663, C747 in human POLE1), a feature unique to Pol e among B-family polymerases (ter Beek et al. 2019; Jain et al. 2014)
- DNA double helix - template and primer backbone tubes with base-pair rungs
- Mutation hotspot markers - somatic EDMs (P286R, V411L, S297F, A456P, P436R, S459F per Leon-Castillo 2020) and germline PPAP variants (L424V, S297F, F367S, M444K, S459F per Palles 2013 / Mur 2023) with interactive tooltips. Note: L424V (Leu->Val) is the germline PPAP founder; somatic L424I (Leu->Ile) is a distinct variant. D287E is included as a somatic recurrence but is not in the Leon-Castillo pathogenic set. 10 variants total in the variant catalogue
- Accessory subunits - ghost representations of POLE2 (p59), POLE3 (p17), POLE4 (p12)
- PCNA sliding clamp - toroidal processivity factor with homotrimer symmetry markers and tripartite Pol e contact interface (PIP-box Q1180, Thumb insertion res. 1102-1122, P-domain contact)
- Beta-sheet hydrogen bonds - inter-strand H-bond visualization (schematic, not from DSSP)
- Exo-Pol shuttling path - animated primer terminus transfer between active sites (~40 A, per Roske & Yeeles 2024). Note: the current animation shows a straight-line path; Roske & Yeeles demonstrate a multi-step DNA handover (Thumb -> P domain [frayed substrate] -> back to Thumb [mismatch excision], with 1-bp backtracking), with alpha-helix 3 of the Exo domain ("wedge helix") physically blocking the direct path. The nascent strand is held throughout by an Anchor loop (~residues 970-980, K974/R975/R976). Replacing the linear tween with a four-state morph across the proofreading intermediates (9F6I->9F6J->9F6K->9F6L) is a planned upgrade
| Mode | Description |
|---|---|
| Domain | Default domain-based coloring |
| B-factor | Approximate flexibility gradient by SS type (not crystallographic B-factors) |
| Conservation | Domain-level evolutionary conservation estimate (not per-residue MSA/ConSurf) |
| Variant | COSMIC/ClinVar variant density (domain-level approximation) |
| Charge | Amino acid formal charge (domain-level average, not APBS electrostatics) |
| Pathogenicity | Clinical significance spectrum (blue-green = benign, white = VUS, red = pathogenic) |
- Default - muted, publication-quality palette
- PyMOL - classic PyMOL-style bright colors
- Accessible - colorblind-friendly palette
A separate page (molstar.html) provides a real coordinate viewer using Mol*, the same viewer used by RCSB PDB and PDBe. Features:
- Loads PDB 9F6D (Roske & Yeeles 2024) with domain-colored cartoon representation
- Finger conformational triad: switch between open (9F6D), ajar (9F6E), and closed (9F6F) states. Four additional mismatch proofreading intermediates (9F6I-9F6L) are deposited but not yet wired into the selector. Construct note: 9F6I and 9F6J use an exonuclease-dead construct (D275A/E277A); active-site labels at the exo site correspond to alanines, not native aspartate/glutamate catalytic residues
- PDB 9B8S (He et al. 2024) — independent Pol e-PCNA cryo-EM structure
- Real DSSP secondary structure, crystallographic B-factors, metal coordination geometry
- Multiple representations: cartoon, ball-and-stick, surface, putty (B-factor), spacefill
- Domain coloring, chain coloring, element coloring, secondary structure coloring
- Toggle Mol* built-in controls for advanced operations
A variant analysis page (variant.html) provides per-mutation analysis for all characterized POLE pathogenic variants:
- ACMG classification with Bayesian posterior probabilities (Tavtigian et al. 2018 likelihood-ratio framework, applied per Mur et al. 2023)
- Auditable evidence codes — each PS/PM/PP code has a literature citation
- Mutational signature attribution — per-variant SBS10a/b/28/14 weights with bar charts
- DDG stability prediction — heuristic destabilization estimate (not FoldX/Rosetta)
- Mol structural context* — variant residue highlighted in PDB 9F6D coordinates
- URL-addressable variants:
variant.html#P286R,variant.html#V411L, etc.
A separate page (mutation.html) visualizes the ultra-rare c.138del germline variant (p.Leu46Phefs*8), showing:
- Side-by-side wild-type vs. truncated mutant comparison
- Domain retention bar chart (stop codon at position 53; 52 / 2,286 residues = 2.3% translated)
- Reading frame analysis with frameshift coloring
- Hypothesis panel (H1–H5) for the observed genotype-phenotype anomaly
Clinical note: c.138del is classified as Pathogenic by the carrier's clinical lab. The carrier presents with a PPAP-consistent phenotype (polyposis, multiple primaries). While truncating POLE variants are not canonical PPAP drivers (PPAP requires an active but proofreading-deficient polymerase), the observed phenotype creates a genotype-phenotype anomaly under active investigation. Five candidate mechanisms are evaluated in the hypothesis panel. Note: the Mur et al. 2023 gene-specific ACMG/AMP framework covers germline variants in the exonuclease domain of POLE/POLD1; c.138del (NTD, LOF) falls outside its scope by both criteria. PVS1 is explicitly excluded from the framework. Biallelic POLE LOF variants cause recessive syndromes (FILS, IMAGe-I) distinct from PPAP.
A Julia data pipeline (pipeline/) computes variant scoring, mutational signature deconvolution, Bayesian variant classification, and structural math from PDB coordinates, outputting JSON consumed by the viewer at runtime. See pipeline/README.md for details. When pipeline JSON is unavailable, the viewer falls back to built-in default values.
Serve the project directory with any static HTTP server:
npx serve .Then open http://localhost:3000 in a modern browser.
| Input | Action |
|---|---|
| Drag | Orbit camera |
| Shift+Drag | Pan |
| Scroll | Zoom |
| Hover | Inspect elements (tooltip) |
1-8 |
Focus domain |
R |
Toggle auto-rotate |
S |
Screenshot (PNG) |
M |
Measure distance |
H |
Toggle helix H-bonds |
B |
Toggle beta-sheet H-bonds |
Z |
Toggle Zn-finger motifs |
P |
Toggle PCNA clamp |
D |
Toggle DNA |
L |
Toggle labels |
E |
Toggle Exo-Pol path |
? |
Keyboard shortcuts |
This viewer is a schematic teaching tool, not a molecular graphics program:
- Coordinates are procedural. Ribbon paths are generated algorithmically to approximate domain topology, not parsed from PDB ATOM records. Domain centers are hand-placed to reflect the spatial arrangement in the Roske & Yeeles 2024 structures but are not superimposed on deposited coordinates.
- B-factors are not crystallographic. The B-factor color mode assigns a single pseudo-value per secondary-structure type (helix = 0.2, strand = 0.5, loop = 0.9). Real B-factors require a refined coordinate set.
- Conservation is domain-level. Values are literature-derived per-domain estimates, not per-residue Shannon entropy from a multiple sequence alignment. For real per-residue conservation, use ConSurf with UniProt Q07864.
- Electrostatic surface is not computed. The charge mode shows formal amino acid charge averaged per domain, not a Poisson-Boltzmann (APBS/PDB2PQR) electrostatic potential mapped onto a solvent-excluded surface.
- The molecular envelope is a geometric approximation, not a Connolly/MSMS solvent-excluded surface.
- H-bonds are schematic. They are placed along the backbone to illustrate i->i+4 and inter-strand patterns, but donor-acceptor distances and angles are not computed from atomic coordinates via DSSP.
- Polymerase active-site numbering is provisional. D640, D642 (Motif A) and D860 (Motif C) are human POLE1 (UniProt Q07864) numbers that have not been independently verified against the deposited 9F6D coordinates. He et al. 2024 (PDB 9B8S) report D626, V627 (Motif A) and D862 (Motif C) as the catalytic residues; this numbering has not been reconciled with 9F6D coordinates in this tool. The literature commonly reports catalytic aspartates in S. cerevisiae Pol2 numbering (D640, D877); the human-yeast offset is ~14 residues in this region but is not constant. Every residue label in this tool carries its organism and numbering source; cross-check against UniProt Q07864 active-site features before citing.
For coordinate-derived molecular visualization, load PDB 9F6D (human Pol e-PCNA-DNA, open Finger conformation) into Mol* or PyMOL.
- UniProt Q07864 (POLE_HUMAN)
- PDB: 9F6D, 9F6E, 9F6F (Roske & Yeeles 2024, matched DNA: open/ajar/closed Finger); 9F6I, 9F6J, 9F6K, 9F6L (Roske & Yeeles 2024, mismatch proofreading intermediates); 9B8S (He et al. 2024); 4M8O (Hogg et al. 2014)
- Roske & Yeeles, Nat Struct Mol Biol 31:1921-1931 (2024) — first human Pol e catalytic domain structures: open (9F6D), ajar (9F6E), and closed (9F6F) Finger conformations; four mismatch proofreading intermediates: post-insertion (9F6I), arrest (9F6J), frayed substrate (9F6K), mismatch excision (9F6L)
- He, Wang, Yao, O'Donnell & Li, Nat Commun 15:7847 (2024) — independent human Pol e-PCNA cryo-EM structure (9B8S)
- Mur, Viana-Errasti et al., Genome Med 15:85 (2023) — gene-specific ACMG/AMP framework for POLE/POLD1 germline variant classification with likelihood-ratio-based Bayesian update
- Yuan, Georgescu, Schauer, O'Donnell & Li, Nat Commun 11:3156 (2020) — structure of the polymerase e holoenzyme and atomic model of the leading strand replisome
- Robinson, Coorens et al., Nat Genet 53:1434 (2021) — normal-tissue somatic mutation rates in POLE/POLD1 germline carriers (Sanger)
- Hogg, Osterman, Bylund, Ganai, Lundstrom, Sauer-Eriksson & Johansson, Nat Struct Mol Biol 21:49-55 (2014) — S. cerevisiae Pol2 catalytic domain (4M8O)
- Palles et al., Nat Genet 45:136-144 (2013) — germline POLE/POLD1 proofreading-domain mutations predispose to colorectal adenomas and carcinomas; defines PPAP and germline L424V
- Leon-Castillo et al., J Pathol 250:323-335 (2020) — POLE-score and 11 pathogenic somatic EDMs for endometrial carcinoma molecular classification
- Parkash et al., Nat Commun 10:373 (2019) — structural consequence of P286R (via yeast P301R); R blocks nascent-strand exo-site access
- ter Beek et al., Nucleic Acids Res 47:5712-5722 (2019) — structural evidence for [4Fe-4S] in the CysX motif of the Pol e catalytic core
- Jain et al., J Mol Biol 426:301-308 (2014) — iron-sulfur cluster in the polymerase domain of yeast Pol e
- Ganai, Bylund & Johansson, Nucleic Acids Res 43:932-942 (2015) — biochemical basis of pol<->exo shuttling
- Alexandrov et al., Nature 578:94-101 (2020) — mutational signatures SBS10a/10b/14/28 (POLE); note COSMIC v3.4 added SBS10c/SBS10d (POLD1-associated)
- Rayner et al., Nat Rev Cancer 16:71-81 (2016) — review of polymerase proofreading domain mutations in cancer
Apache 2.0 — see SECURITY.md for vulnerability reporting.