Source 1primary paper2026MED
Toolkit/TripleMatcher
TripleMatcher
Taxonomy: Technique Branch / Method. Workflows sit above the mechanism and technique branches rather than replacing them.
Summary
we introduce TripleMatcher, which searches for a triple-helix pattern, filters candidates by C1'-C1' distance thresholds, and merges overlaps into region-level zones
Usefulness & Problems
No literature-backed usefulness or problem-fit explainer has been materialized for this record yet.
Published Workflows
Objective: Annotate and detect RNA triple helices from sequence and secondary structure, then reduce raw candidates to a small interpretable set suitable for targeted experimental validation.
Why it works: The workflow first represents Hoogsteen contacts explicitly, then searches for matching triple-helix patterns, and finally removes geometrically implausible candidates using C1'-C1' distance thresholds. The abstract states that better secondary-structure inputs, especially from pseudoknot-aware predictors, align with improved downstream triple-helix recovery.
encoding Hoogsteen contacts in secondary-structure annotationpattern-based triple-helix searchgeometric filtering by C1'-C1' distance thresholdsoverlap merging into region-level zonessecondary-structure annotationcomputational pattern matchinggeometric candidate filteringpredictor benchmarkinglarge-scale computational screening
Stages
- 1.Hoogsteen-aware secondary-structure annotation(library_design)
This stage creates a secondary-structure representation that explicitly includes Hoogsteen contacts needed for triple-helix annotation and downstream search.
Selection: encode Hoogsteen contacts in an extended dot-bracket representation
- 2.Triple-helix pattern search(broad_screen)
This stage generates raw candidate triple-helix regions from the annotated or predicted secondary structures.
Selection: search for a triple-helix pattern in sequence and secondary-structure input
- 3.Geometric feasibility filtering(counter_screen)
This stage removes geometrically implausible candidates and improves precision and F1 while maintaining sensitivity.
- 4.Region-level consolidation(hit_picking)
This stage consolidates overlapping candidates into region-level zones to produce a smaller interpretable output set.
Selection: merge overlaps into region-level zones
- 5.Secondary-structure predictor benchmarking(secondary_characterization)
This stage evaluates which upstream structure-prediction methods best support the detection framework.
Selection: compare eight predictors for their ability to reproduce local architecture required for detection
Steps
- 1.Extend dot-bracket notation with a third annotation lineannotation scheme
Represent Hoogsteen contacts needed for RNA triple-helix annotation.
The search framework requires a representation that explicitly captures triple-helix contacts before pattern matching can be applied.
- 2.Search for triple-helix patterns with TripleMatchercomputational detector
Generate raw candidate triple-helix regions from sequence and secondary-structure input.
Broad candidate generation occurs before geometric filtering so that potentially valid regions are not excluded prematurely.
- 3.Filter raw candidates by C1'-C1' distance thresholdsfiltering method
Remove geometrically implausible and spurious candidates while maintaining sensitivity.
This lower-cost in silico filter narrows the broad candidate set before interpretation or targeted validation.
- 4.Merge overlapping candidates into region-level zonespost-processing method
Convert overlapping candidate calls into a smaller interpretable set of regions.
Consolidation is performed after filtering so the final output is compact and suitable for targeted experimental validation.
Taxonomy & Function
Primary hierarchy
Technique Branch
Method: A concrete computational method used to design, rank, or analyze an engineered system.
Target processes
recombinationselectionValidation
Cell-freeBacteriaMammalianMouseHumanTherapeuticIndep. Replication
Supporting Sources
Ranked Claims
Using 8 RNAs with experimentally established triple helices, TripleMatcher localized all annotated regions and geometric filtering improved precision and F1 while maintaining sensitivity.
F1 0.42 to 0.62positive predictive value 0.42 to 0.81structure-wise detection 8/8
TripleMatcher searches for triple-helix patterns, filters candidates by C1'-C1' distance thresholds, and merges overlaps into region-level zones.
The paper presents a secondary-structure-based framework to annotate and detect RNA triple helices.
Applied prospectively, the framework scaled to a screen of 4160 RNAs and distance filtering reduced 150990 raw candidates to 97 geometrically feasible regions across seven molecules.
geometrically feasible regions 97median raw candidates per molecule 108molecules with feasible regions 7raw candidates 150990RNAs screened 4160
Approval Evidence
1 source4 linked approval claimsfirst-pass slug triplematcher
we introduce TripleMatcher, which searches for a triple-helix pattern, filters candidates by C1'-C1' distance thresholds, and merges overlaps into region-level zones
Source:
benchmark performancesupports
Using 8 RNAs with experimentally established triple helices, TripleMatcher localized all annotated regions and geometric filtering improved precision and F1 while maintaining sensitivity.
Source:
method capabilitysupports
TripleMatcher searches for triple-helix patterns, filters candidates by C1'-C1' distance thresholds, and merges overlaps into region-level zones.
Source:
method introductionsupports
The paper presents a secondary-structure-based framework to annotate and detect RNA triple helices.
Source:
screening scalesupports
Applied prospectively, the framework scaled to a screen of 4160 RNAs and distance filtering reduced 150990 raw candidates to 97 geometrically feasible regions across seven molecules.
Source:
Comparisons
No literature-backed comparison notes have been materialized for this record yet.
Ranked Citations
- 1.