AptaFold

/ Introduction

What it does.

• Four-cation aware folding

  Explicit Na⁺, Mg²⁺, K⁺, Ca²⁺ inputs with temperature — the only open tool we know of that touches Mg²⁺-vs-Ca²⁺ for DNA aptamers.

• G-quadruplex detection & cartoon

  Detects G-tract motifs, scores topology under K⁺ / Na⁺ channel binding, and renders both axial (top-down tetrad) and side (layer-diagram) views via a DrawTetrado-style painter's algorithm. The main fold viewer also auto-switches to 3D when a G4 is found, with a Flat ↔ 3D toggle.

• Four interchangeable engines

  seqfold (pure-Python fallback), ViennaRNA (industry default, monovalent salt), ViennaRNA+Mg²⁺ (experimental Mg²⁺ callback), and 4cations DP (full four-cation experimental engine).

• Suboptimal-fold exploration

  Wuchty-style enumeration of folds within ΔE of the MFE, with a coverage cap and "show 10/20/30" controls — useful when the MFE alone hides the relevant conformation.

• Motif-aware constraints

  "Loose scaffold (SELEX)" mode keeps the known fixed primer-binding region as scaffold and lets the random N region fold freely — a one-click constraint that matches how SELEX libraries are actually designed.

• Multi-format export

  PDF, PNG, SVG, JSON, CSV, and PPTX. Also includes a batch mode for running many sequences at once through the same parameter set.

/ Engines

• seqfoldPure-Python fallback. No salt model. Useful when ViennaRNA isn't available.
• ViennaRNAIndustry-default thermodynamic engine with Lai et al. 2023 monovalent salt correction.
• ViennaRNA+Mg²⁺Experimental: ViennaRNA + per-loop Mg²⁺ callback (uncalibrated).
• 4cations DPExperimental: full four-cation dynamic-programming engine (uncalibrated).

/ Usage

How to use it.

1. 01

   Download & launch

   Click the download button above (AptaFold-0.6.7-onefile.exe, 114 MB). Just double-click the downloaded EXE — it's a self-contained single-file build, no install step, no folder to manage. The main window opens straight up and shows the sequence box, ionic-condition sliders, engine picker, and fold buttons. First-run takes a couple of extra seconds while the bundle unpacks to a temp folder.

   

   Screenshot · 01-main-window.png

2. 02

   Enter sequence & ionic conditions

   Type or paste the sequence (ATCG / AUCG). Set [Na⁺], [Mg²⁺], [K⁺], [Ca²⁺] in mol/L and the temperature in °C — defaults match a common SELEX selection buffer. Choose an engine (default 4cations), and tick options: G-quad scoring, Suboptimals within ΔE, Loose scaffold (SELEX) for motif-aware constraints.

3. 03

   Fold & read the result

   Click Fold. The result panel renders the secondary structure inline — base-pair stems drawn with Watson–Crick H-bonds in red, with position numbers, 5′ / 3′ termini, and ΔG values printed alongside each fold. You can Re-fold after changing conditions without reloading anything.

   

   Screenshot · 02-fold-hairpin.png

   When the predicted fold contains a G-tetrad, AptaFold detects the G4 and defaults the result viewer to the 3D oblique-cube layout — the four G-tracts stack as parallel rails so the tetrad geometry reads at a glance. Click the Flat ↔ 3D toggle in the bottom-left of the result panel to switch to the conventional flat secondary-structure layout (and back). The toggle is sticky per fold, so a re-fold preserves your view choice.

   

4. 04

   G-quadruplex detail (if found)

   If G-tract motifs are detected (and G-quad scoring is on), the G-quad… button activates. Open it for a layered cartoon: axial view shows the tetrad from above with the central cation, side view shows tetrad stacking with propeller / lateral / diagonal loops drawn as external arcs. Each view exports as PNG / SVG / JSON / CSV.

   

   Fig. 03/04 — G-quadruplex side & axial

5. 05

   Export, or run a batch

   Export… writes the current fold (or the whole suboptimal set) as PDF, PNG, SVG, JSON, CSV, or PPTX. For many sequences at once, switch the Input source from Manual sequence to a CSV / Excel / Liu-Lab Aptamer Analysis output — AptaFold runs the same parameter set across all rows and exports per-sequence files.

/ Disclaimer

Things to know.

1. Engineering approximation, not calibrated physical chemistry.

   The four-cation engine combines published trends (ViennaRNA monovalent salt correction, Tan–Chen tightly-bound-ion ideas, G-quadruplex channel-binding) into a single model. It is intentionally conservative and uncalibrated — use the output as a trend check and starting hypothesis, not as a quantitative T_m prediction.

2. Mg²⁺ / Ca²⁺ effects are experimental.

   No open folding tool natively handles divalent cations — ViennaRNA's salt correction is monovalent-only. AptaFold's ViennaRNA+Mg²⁺ and 4cations DP engines are experimental and pending the wet-lab T_m / ITC calibration described in the project's EXPERIMENT_PLAN.md.

3. Local processing, no telemetry.

   AptaFold makes no network calls. Sequence input, folding, rendering, and exports all run locally. There is no auto-update check, no telemetry, no analytics.

4. MIT license · academic & non-commercial use.

   AptaFold is released under the MIT license. Free for academic research, teaching, and personal use. ViennaRNA itself ships under its own permissive licence — see the bundled licenses/ notice.

5. Provided as-is, without warranty.

   Fold output is best-effort. Validate critical findings with independent methods — m-fold / UNAFold for Mg²⁺ comparison, CD / UV-melting for G-quadruplex topology. Bug reports: [email protected].

6. Cite if used in publications.

   If AptaFold contributed to a published study, please cite: Zhang X. AptaFold (dna-fold), v0.6.7, Liu Lab, University of Waterloo, 2026. aptapilot.com/tools/aptafold.

/ Acknowledgments

Built on prior work.

• 

  Prof. Juewen Liu

  Training data & support

  Professor of Chemistry at the University of Waterloo. Tier 1 Canada Research Chair in Biosensors & Bionanotechnology. Provided the invaluable wet-lab training data and ongoing support that made the four-cation calibration possible.

• 

  Xiaohan Zhang

  Author · Development · Formulas · ML

  Postdoctoral researcher at the University of Waterloo under Prof. Juewen Liu. Main developer — designed and implemented the engine, derived the formulas, tuned the parameters, and built the machine-learning components of the four-cation model.

• 

  Yuanli Li

  Data collection & cleaning

  Third-year undergraduate in the Materials Nanoscience joint program (Beijing Jiaotong University × University of Waterloo). Built the training dataset — collected the literature and wet-lab measurements, cleaned them into the calibration-ready form used by the four-cation engine.

1. ViennaRNA · Lorenz, Hofacker, et al.

   The industry-default RNA folding engine. AptaFold's default and +Mg²⁺ engines are both layered on top of ViennaRNA, including the Lai et al. 2023 monovalent salt correction.

2. DrawTetrado · Zok et al. 2022

   AptaFold's oblique G4 cube uses a painter's-algorithm depth ordering and propeller-loop arc style directly inspired by DrawTetrado (Zok et al., Bioinformatics 38:3835, 2022) — the purpose-built G-quadruplex layer-diagram tool that nailed the visual conventions.

3. Tan, Chen et al. — tightly-bound-ion model

   The per-loop divalent callback design draws on the Tan–Chen tightly-bound-ion framework — the most physically principled treatment of Mg²⁺ in nucleic-acid folding, even though it had never been integrated into a distributable folding tool before.

• ViennaRNA
• seqfold
• PySide6
• NumPy
• matplotlib
• python-pptx
• openpyxl
• pypdf
• xlrd
• numba

/ FAQ

Questions people actually ask.

1. How is this different from m-fold / UNAFold?

   AptaFold takes the ionic conditions of an aptamer assay as explicit inputs. Beyond ViennaRNA's monovalent salt correction it adds a per-loop divalent callback for Mg²⁺ and Ca²⁺, and a G-quadruplex channel-binding term for K⁺ / Na⁺. It also exposes every folding parameter in the GUI rather than hiding them behind a CLI.

2. Why do salt conditions change the predicted structure?

   Duplex and loop stabilities are ion-dependent. Divalent Mg²⁺ and Ca²⁺ stabilise folded states differently from monovalent Na⁺, and K⁺ specifically stabilises G-quadruplex channels. Folding at default 1 M Na⁺ when your binding buffer is, say, 100 mM Na⁺ with 2 mM Mg²⁺ can return a different minimum-free-energy structure than the one your assay actually samples.

3. Does it detect and draw G-quadruplexes?

   Yes. G-quadruplex detection is built in, and topologies are rendered as a layered cartoon view rather than left for you to read off a dot-bracket string or a paper figure.

4. Can I constrain folding to a known motif?

   Yes — motif-aware constraints are supported through the Loose-scaffold SELEX mode, which is intended for libraries built around a fixed scaffold with randomised regions.

5. What platform does it run on, and what licence?

   AptaFold is a Windows desktop application, installed and kept current through the AptaPilot Launcher. It is released under the MIT licence.

/ Citation

How to cite.

If this tool contributed to work you are publishing, a citation is the most useful thing you can send back — it is what keeps the tool maintained and lets other people find it.

@software{zhang_aptafold_2026,
  author  = {Zhang, Xiaohan},
  title   = {AptaFold},
  year    = {2026},
  version = {0.6.7},
  url     = {https://aptapilot.com/tools/aptafold/}
}

/ Also in the toolkit

Other tools.

Desktop · Windows + macOS

Aptamer Analysis Tool

SELEX FASTQ trim, family clustering, and PDF / PNG / XLSX / Word reports. AptaFold ingests its output as a batch source.

Web · Online

K_d Fitting Tool

Quadratic binding-model K_d fitting from concentration / signal data. Browser-only, no upload.

Desktop · Windows · Experimental

Structural Family Classification

Group SELEX aptamers by loop-region structure with a self-written motif-block classifier. Reuses AptaFold for folding.