Paper Reproductions — 63/63 Papers in Sovereign Rust¶
wetSpring reproduced results from 63 peer-reviewed papers across 5 research groups and 4 tracks. Every reproduction has:
- A Python/R baseline from the original methodology
- A Rust implementation with quantitative parity checks
- A GPU validation tier (50/50 three-tier eligible)
- Full provenance chains via NUCLEUS composition
This notebook maps the evidence across researchers, departments, and domains.
For other springs: create your own researcher/paper map. The structure is: one section per PI, one row per experiment, frozen JSON for the numbers.
Researcher Map¶
| Researcher | Department | Institution | Domain | Papers |
|---|---|---|---|---|
| Christopher Waters | MMG | Michigan State | Quorum sensing, c-di-GMP | 15+ |
| Kevin Liu | CMSE | Michigan State | Comparative genomics, phylogenetics | 10+ |
| Jesse Cahill & Chuck Smallwood | Bioscience | Sandia National Labs | Biosurveillance | 5+ |
| A. Daniel Jones | BMB/Chemistry | Michigan State | Mass spectrometry, PFAS | 8+ |
| Rika Anderson | Biology | Carleton College | Vent metagenomics, pangenomics | 5+ |
Track 1: 16S Metagenomics (Waters, Anderson)¶
The core pipeline — FASTQ to taxonomy to diversity. Galaxy/QIIME2 replaced entirely with sovereign Rust. 30 bio modules, 1 dependency.
Experiment 001: Galaxy Bootstrap Checks: 8/8 ASVs: 232 Phyla: 9 Status: PASS R/vegan Cross-Validation (Exp 335): Tool: vegan v2.7.3 shannon_uniform_10: 2.302585092994045 simpson_uniform_10: 0.9 chao1_estimate: 20 pielou_uniform: 0.9999999999999998 bray_curtis_ab: 0.4 Rarefaction monotonic: True
Track 2: Analytical Chemistry (Jones)¶
LC-MS feature extraction (Asari), PFAS screening, VOC biomarkers, spectral cosine matching.
Track 2: 8/8 checks PASS
Runtime: 16.4s
exp005_asari:
Passed: 5/5
Runtime: 15.4s
features: 5,951
compounds: 4,101
exp006_findpfas:
Passed: 3/3
Runtime: 0.77s
candidates: 62
unique: 25
Paper-extracted benchmarks:
Carney2016 Crash Agents: 5 entries
Humphrey2023 Metrics: 5 entries
Reese2019 Voc Biomarkers: 5 entries
Reichardt2020 Spectroradiometric: 5 entries
Track 3: Phylogenetics & Comparative Genomics (Liu)¶
Tree reconstruction, bootstrap support, ancestral state reconstruction, HMM-based gene family analysis.
Track 3: Phylogenetics & Comparative Genomics Experiment Files ----------------------------------- 019 Phylogenetic 1 021 RF Baseline 1 022 Gillespie 1 026 HMM 1 028 Alignment 1 029 Felsenstein 1 031 Bootstrap 1 032 Placement 1 036 PhyNetPy RF 1 037 PhyloHMM 1 038 SATE Pipeline 1 Total Track 3 experiments with data: 11
Track 4: Soil & Environmental (Anderson + QS)¶
Anderson localization applied to soil microbial ecology — the key scientific discovery bridging wetSpring to groundSpring. 9 soil experiments validated against published field studies.
Track 4: 9 soil experiments with frozen baselines 170 Soil QS Pore Geometry (martinez2023): 3 data sections 171 Soil Pore Diversity (feng2024): 6 data sections 172 Distance Colonization (mukherjee2024): 5 data sections 173 No-Till Brandt Farm (islam2014): 5 data sections 174 No-Till Meta-Analysis (zuber2016): 4 data sections 175 Long-Term Tillage (liang2015): 3 data sections 176 Biofilm Aggregate (tecon2017): 4 data sections 177 Structure-Function (rabot2018): 3 data sections 178 Tillage Microbiome (wang2025): 4 data sections
Aggregate Evidence¶
Total experiment directories: 54 Directories with frozen data: 54
Key Discovery: Three-Tier Validation¶
wetSpring pioneered the pattern now used across all 8 springs:
Tier 1: Python/R baseline → frozen JSON
Tier 2: Rust implementation → [OK]/[FAIL] parity checks
Tier 3: GPU (barraCuda) → tolerance-checked acceleration
50 of 50 three-tier eligible papers have complete CPU + GPU + metalForge validation. The remaining 13 papers are CPU-only (no GPU math component).
baseCamp Papers: 01, 03, 04, 05, 06 on primals.eco/science
Faculty Briefings: whitePaper/baseCamp/*.md in the wetSpring repository
Source: syntheticChemistry/wetSpring