G-QUADRUPLEX ANALYSIS

Predict G-quadruplex motifs.

Find potential G4 motifs, compare method-specific scores, and inspect guanine and GC composition across your sequences.

● Local sequence analysis

01Sequence input

Nucleotide sequence or multi-record FASTA

DNA or RNA · IUPAC ambiguity preserved · Coordinates remain relative to each input record.

Add a sequence or load the example to begin.

02Prediction methods

3 selected

Select methods

Search parameters

Search strands

Minimum G-tract

Maximum G-tract

Minimum loop

Maximum loop

Maximum tract-motif length

G4Hunter window

G4Hunter threshold

Allow one bulge or mismatch in G-tracts

Unchecked: selected tract lengths mean consecutive guanines for every tract-based method, including QGRS. ImGQ single-defect search is skipped in strict mode.

Tract and loop settings apply to all motif-search engines. QGRS and g4predict-style calls use equal tract lengths within each candidate; the selected range is searched one length at a time. G4Hunter reports scored regions without assigning four G-tracts; its region lengths and guanine runs can differ. QGRS-C++ uses its documented scoring convention.

Methods not integrated

Original QGRS Mapper web scoring is not integrated. Other original runtimes or trained model weights are not loaded in this browser app. These methods are skipped; no substitute score is returned.

pqsfinder · QPARSE · G4Catchall · G4-iM Grinder · Quadron · G4RNA screener/G4NN · PENGUINN · G4detector · G4Boost · rG4detector · G4mismatch · DeepG4 · G4Beacon · epiG4NN · eG4finder · DeepGQ · G4mer · TransG4 · G4-Attention · iDualG4 · G4STAB · G4ShapePredictor

Sequences are analyzed in your browser.
NCBI is contacted only for record retrieval.

Prediction results

Explore motif calls and their sequence-level properties.

Excel preserves color and underline. CSV is plain text. All exports include every filtered call.

—sequence records

—nucleotides analyzed

—method-specific motif calls

—methods run

Your sequence, resolved into candidates.

Run a prediction to inspect G4 motifs, strand positions, and scores.

Each method retains its own calls. Overlapping calls are not independent structures.

Read the scores correctly

Tool score follows the documented scoring rule of the selected browser implementation. QGRS-C++ is a separate score convention from the original QGRS Mapper. Quadparser, TetraplexFinder-style search and ImGQ-style search report motif matches without a calibrated score.

G % = 100 × G / known-base count. GC % = 100 × (G + C) / known-base count. Only A/C/G/T contribute to this denominator; ambiguous bases are excluded and counted separately in the motif details.

G4Hunter mean assigns each base in a G-run of length r the value +min(r,4), and each base in a C-run −min(r,4); other bases are zero. The sequence mean is their sum divided by sequence length. Detection uses full-sequence window scores; the final merged, refined region is rescored.

g4predict-style score = 40 × tetrads − (loop + bulge nucleotides) − 10 × bulges. Every candidate uses equal-length G-tracts. Strict mode requires consecutive G-tracts. Enabling the imperfection option allows one internal single-base bulge for g4predict-style search and a single bulge or mismatch for ImGQ-style search. Vacancy mode is not implemented.

cG/cC counts overlapping G and C substrings of every length i, weighted by 10i. A maximal run of length r contributes 10r(r+1)(r+2)/6 to its component. If cC is zero, the denominator is set to 1, following the G4RNA screener implementation. This app scores the reported sequence only; neighboring RNA context is not added automatically.

QGRS-C++ G-score definition

floor((M − 4t − 1)(t − 1) − D), where D = (|L1−L2| + |L2−L3| + |L1−L3|)/3; t is tetrad count. The documented C++ convention uses M=30 for two tetrads and M=45 otherwise. The column reports the highest score across supported whole-interval canonical decompositions satisfying the selected tract and loop settings. No score is assigned when no supported decomposition exists.

QGRS C++ implementation ↗

Methods and reproducibility

Quadparser-style returns one greedy match per eligible start. QGRS-style retains the best conformation per start and tetrad count. TetraplexFinder-style uses canonical perfect tracts; its native bulge mode is not included.

Scoring audit · v1.3

The formulas were checked against the original publications and author-maintained code linked below. G4Hunter, cG/cC, G4P, G % and GC % retain their verified arithmetic. The g4predict perfect-tract search now enumerates equal tetrad counts, correcting scores when a range of tract lengths is selected. Each g4predict interval retains the highest-scoring interpretation among the perfect and single-bulge candidates searched.

QGRS-C++ is not original QGRS Mapper. For the documented example GGGGTGGGGTGGGGTGGGG, the original Mapper help reports 63 under its default 30-nt scoring convention, while this C++ convention returns 84 with its fixed 45-nt convention for four tetrads. Do not transfer Mapper thresholds or the review’s benchmark performance matrix to this app.

The previous software checks verified code and exports; they did not establish equivalence to every original server. Motif enumeration and overlap handling remain the explicitly described browser implementations.

The available engines are independent browser implementations of documented rules, with their scope identified in the selector. They do not invoke the original web servers or trained machine-learning models.

References & implementation scope

Puig Lombardi & Londoño-Vallejo, 2020 review: method matrix and benchmark definitions.
Original QGRS Mapper score documentation: separate from the implemented C++ variant.
G4RNA screener cG/cC reference implementation: overlapping counts and zero-C convention.
Quadparser: canonical four-tract motifs.
QGRS Mapper: motif enumeration; scoring uses the explicitly named C++ variant.
G4Hunter: sliding-window G/C run scoring.
TetraplexFinder: configurable canonical motif rules.
g4predict: tetrad reward with loop/bulge penalties.
ImGQfinder: single-imperfection rule family.
G4P Calculator: regional window-density report.
NCBI E-utilities: identifier resolution and sequence retrieval.

A prediction identifies a candidate sequence under the method’s assumptions. It does not establish formation in cells or a particular three-dimensional fold.