🎓 Guided Walkthrough • Step 1 of 4
Stage 0: Alignment, Taxon & Site Coherence QA
BUSHWHACKER // 64-SITE BIT-PLANE SIMD WEBASSEMBLY WORKBENCH
Parameter-Free Phylogenetics, Polytomy Calibration & Shunt Discovery via MDL
Drop any DNA/RNA, 20-state Protein, or Binary / Discrete / Copy-Number character matrix (FASTA, FAA, PHYLIP, NEXUS, CSV/TSV), Newick tree, or lossless .bwz archive. BUSHWHACKER replaces continuous branch-length priors with exact combinadic prefix codes — automatically contracting unsupported splits into calibrated polytomies and peeling convergent or horizontal gene transfer shunts 100% locally in your browser.
Drop DNA, Protein, Binary/CSV Matrix, Newick, or .bwz
Accepts
.fasta, .faa, .phy, .nex, .csv/.tsv (`0/1` or `K`-state), .nwk, .json, or lossless .bwz
100% Client-Side Rust WebAssembly • Zero Data Leaves Your Machine
🎓 First-Time User Walkthrough
~60 sec interactive guided tour • Live MDL math inspector
How BUSHWHACKER Works: 4-Stage Interactive Guided Tour
Launch a guided tour below — a live coach banner will explain each stage and provide one-click forensic actions as you move from raw alignment QA to polytomy calibration and reticulate shunt discovery.
Stage 0 • Alignment & Site QA
Dual-Entropy Site Coherence
Audits taxon gap/ambiguity load, deduplicates identical sequences, and computes 3rd-codon-position entropy weights (
L_eff) and the exact Star-Tree null code length.Stage 1 • MDL Tree & Polytomies
Combinadic Split Selection
Induces the maximum-compression tree via 64-site bit-plane SIMD, retaining only splits whose parsimony savings strictly beat their topological + combinadic header cost.
Stage 2 • Reticulate Shunt Peeling
SPARSE Convergence & HGT
Peels residual non-tree homoplasies into supported
SPARSE convergent coding shunts or contiguous INTERVAL recombination/HGT tracts with exact codon forensics.Stage 3 • Dossier & Lossless .bwz
Publication Report & Archive
Generates a standalone HTML dossier, annotated NEXUS/Newick trees, and bit-for-bit lossless
.bwz phylogenetic arithmetic-coded archives.Curated Empirical & Synthetic Benchmark Cohorts
0 QA & Site Coherence → 1 MDL Tree & Polytomies → 2 Reticulate Shunts → 3 Dossier & .bwz How it works →
Inspecting alignment & site coherence…
Admitted Taxa
—
Active sequences
Alignment Length
—
Columns & reading frame
Effective Sites (L_eff)
—
Parsimony-informative
Star-Tree Null MDL
—
0-split baseline cost
Data Regime
—
Format & alphabet
| Use | # | Taxon Label & Species / Clade Annotation | Ungapped Len | Gap / Missing % | GC % | Private Mutations (Terminal Load) | Status |
|---|
Stage 1 • MDL Tree & Polytomies
Inducing minimum description length phylogeny…
Tree MDL Code Length
—
L(T*, D) in nats
Net MDL Savings
—
vs. Star Tree Null
Resolved Binary Splits
—
Supported internal stems
Calibrated Polytomy Hubs
—
Zero-evidence contractions
Compute Runtime
—
64-site bit-plane WASM
Stage 2 • Reticulate Shunt Peeling
Scanning residual homoplasies for convergent & HGT shunts…
Supported Shunts
—
Beat combinadic mask tax
INTERVAL Mosaic Tracts
—
Contiguous HGT / recomb
SPARSE Convergent Shunts
—
Parallel coding adaptation
Network MDL Gain
—
Nats saved beyond tree
Top Shunt P_local
—
Conditional local P_local
Self-Contained Publication Dossier & Lossless Archive