Hierarchical Adaptive 'RT-QuIC' Classification for Complex Matrices
Overview
Extends 'RT-QuIC' (Real-Time Quaking-Induced Conversion) statistical analysis to complex environmental matrices through hierarchical adaptive classification. 'KWELA' is named after a deity of the Fore people of Papua New Guinea, among whom Kuru, a notable human prion disease, was identified. Implements a 6-layer architecture: hard gate biological constraints, per-well adaptive scoring, separation-aware combination, Youden-optimized cutoffs, replicate consensus, and matrix instability detection. Features dual-mode operation (diagnostic/research), auto-profile selection (Standard/Sensitive/Matrix-Robust), RAF integration for artifact detection, matrix-aware baseline correction, and multiple consensus rules. Methods include energy distance (Szekely and Rizzo (2013) doi:10.1016/j.jspi.2013.03.018), CRPS (Gneiting and Raftery (2007) doi:10.1198/016214506000001437), SSMD (Zhang (2007) doi:10.1016/j.ygeno.2007.01.005), and Jensen-Shannon divergence (Lin (1991) doi:10.1109/18.61115). This package implements methodology currently under peer review; please contact the author before publication using this approach. Development followed an iterative human-machine collaboration where all algorithmic design, statistical methodologies, and biological validation logic were conceptualized, tested, and iteratively refined by Richard A. Feiss through repeated cycles of running experimental data, evaluating analytical outputs, and selecting among candidate algorithms and approaches. AI systems ('Anthropic Claude' and 'OpenAI GPT') served as coding assistants and analytical sounding boards under continuous human direction. The selection of statistical methods, evaluation of biological plausibility, and all final methodology decisions were made by the human author. AI systems did not independently originate algorithms, statistical approaches, or scientific methodologies.
Initial CRAN Release
6-Layer Hierarchical Architecture
- **Layer 1: Hard Gate** - Biological constraint filter with stochastic rescue (research mode only)
- **Layer 2: Per-Well Scoring** - Profile-dependent adaptive transforms
- **Layer 3: Adaptive Combination** - Separation-aware score combiner
- **Layer 4: Adaptive Cutoff** - Youden-optimized threshold per plate/group
- **Layer 5: Replicate Consensus** - Treatment-level classification
- **Layer 6: Instability Detection** - Matrix interference override
Dual-Mode Operation
mode = "diagnostic"(default): Deterministic classification from TTT/MP/RAF evidence only. No stochastic rescue, no stochastic score blending.mode = "research": Full adaptive architecture with stochastic rescue and distributional scoring.
Instability Detection
Show all changes in 1.0.0Hide the rest of the changes in 1.0.0
compute_instability_flags(): 6 deterministic metrics per treatment (Fano deviation, crossing variability, Wasserstein/Energy distance from both controls, TTT dispersion, RAF-MP inconsistency)- Configurable strictness:
"moderate"(2+ flags),"strict"(1+),"lenient"(3+) - Override:
INCONCLUSIVE_MATRIX_EFFECTwhen minimum flags exceeded
Auto-Profile Selection
profile = "auto"selects based on Cohen's d separation quality
Group-Aware Mixed-Assay Support
- Per-group PC/NC distributions, stats, thresholds, and RAF parameters
- Enables mixed RT-QuIC + Nano-QuIC under single
kwela_analyze()call
RAF Integration
- Artifact detection via RAF/MP ratio analysis
- RAF concordance scoring for quality assessment
Replicate Consensus Rules
"strict","majority"(default),"flexible","threshold"
Stochastic Metrics Library
- Energy distance, Wasserstein distance, Jensen-Shannon divergence
- CRPS, Dawid-Sebastiani, interval score
- SSMD, Fano factor, coefficient of variation
- Cohen's d separation assessment
From NEWS.md in the release
Install
Health
- OK13 OK2026-08-05
- NOTE12 OK, 1 NOTE2026-08-01
- OK14 OK2026-03-10
Documentation
- Help pages
- 12includes a package overview page
- Examples that run
- 73%8 run in a plain check · 3 only when donttest examples are enabled (R CMD check --as-cran)
- Exported functions with examples
- 100%
- Documented parameters
- 96%
- Return-value docs
- 100%
- Help pages citing references
- 33%
Downloads
Repository
Checks run against github.com/rfeissiv/kwela on 2026-10-21.
No development-tooling practices detected in the upstream repository.
Dependencies
Installing KWELA adds no packages beyond those that come with R.
Code & Tests
People & History
- RR 4.6.0 released · 2026-04-24
- 1.0.0Latest2026-02-28 · current release
- RR 4.5.0 released · 2025-04-11
Package metadata
- First published
- 2026-02-28
- Total releases
- 1 / 1 yrs
- License
- MIT + file LICENSE OSI
- Minimum R
- ≥ 4.0.0
- Download size
- 34 KB
Cite
Cite this package
Run in R for the authors' preferred citation:
citation("KWELA")This package has no citation file, so citation() returns this entry built from DESCRIPTION.
Cite the R Observatory
For a number measured here: a download total, a coverage figure, an archival date.
From data release v2026-10-25, which the citation names so these numbers can be found later. More on citing and the projects behind them.