From: ecoPrimal — human + synthetic intelligence
Organization: 🔧🦎 ecoPrimals Date: March 17, 2026 License: All code AGPL-3.0-or-later; all documentation CC-BY-SA-4.0 Repositories: Springs at github.com/syntheticChemistry · Primals at github.com/ecoPrimals · Products at github.com/sporeGarden
What This Is
🔧🦎 ecoPrimals is a sovereign scientific computing ecosystem built in pure Rust with GPU acceleration via WebGPU (WGSL shaders). It replaces Python/R/Fortran/Java tool chains across life science, pharmacology, physics, and data provenance domains. Every claim below has a validation binary that proves it — clone the repo, run the binary, verify the output.
This document provides honest parity assessments: what we match, what we exceed, what proprietary tools still do better, and where to find everything.
What It Replaces
| Commercial/Open Tool | Annual Cost | What It Does |
|---|
| Galaxy | Free (hosted) / $50K+ (local) | Web-based bioinformatics workflow |
| QIIME2 | Free | 16S/ITS amplicon analysis (Python) |
| mothur | Free | 16S OTU-based pipeline (C++) |
| DADA2 (R) | Free | Amplicon sequence variant denoising |
| phyloseq (R) | Free | Microbiome statistical analysis |
| vegan (R) | Free | Community ecology analysis |
ecoPrimals Replacement: wetSpring
Repository: github.com/syntheticChemistry/wetSpring Status: V127 — 1,443+ tests, 306 validation binaries, 376 experiments, 5,707+ checks
| Capability | Parity Level | Notes |
|---|
| FASTQ parsing + quality filtering | Full parity | Sovereign parser, no needletail dependency |
| Read merging (paired-end) | Full parity | Overlap detection, quality-aware consensus |
| Dereplication | Full parity | Hash-based, GPU-accelerated |
| DADA2 denoising | Full parity | Error model + denoising validated against R DADA2 |
| 🦁🐍 Chimera detection | Full parity | de novo + reference-based |
| Taxonomy classification | Full parity | Naïve Bayes, k-mer, spectral matching |
| UniFrac (weighted/unweighted) | Full parity | GPU-accelerated, validated against phyloseq |
| Diversity indices (Shannon, Simpson, Chao1, Pielou) | Full parity | Validated to 1e-12 against textbook definitions |
| Bray-Curtis dissimilarity | Full parity | GPU BrayCurtisF64 kernel |
| PCoA ordination | Full parity | GPU BatchedEighGpu eigendecomposition |
| Rarefaction curves | Full parity | Monte Carlo with bootstrap CI |
| Smith-Waterman alignment | Full parity | Affine gap penalties |
| HMM phylogenetics | Full parity | Forward algorithm, GPU batch |
| Newick tree parsing / Robinson-Foulds | Full parity | Full phylogenetic tree comparison |
| ODE ecological models (Lotka-Volterra, QS, etc.) | Exceeds | 5 biological ODE systems with GPU shader generation |
| Anderson localization for community structure | No equivalent | Novel physics framework — no proprietary tool does this |
| GPU acceleration (all operations) | Exceeds | 150+ GPU primitives, zero local WGSL |
| Spectral cosine matching | Exceeds | 1,077× speedup over CPU |
| Capability | Gap | Path to Parity |
|---|
| GUI workflow builder | No GUI — CLI + validation binaries only | 🌸👅 petalTongue provides visualization; Galaxy-style builder not planned |
| Plugin ecosystem | No third-party plugin system | IPC capability discovery enables composition |
| Training documentation | No tutorials, no workshops | K-Nome methodology document exists; formal curriculum pending |
| Multi-user web interface | Single-user, local execution | 🌿🖥️ biomeOS IPC enables multi-client; web UI not planned |
| Established community | One developer, public repos | 3.2M lines of Rust, 107K+ tests, all validation executable |
| Cloud deployment | Local/LAN only | ⚛️🧬 NUCLEUS bonding model supports distributed deployment |
Where to Find / Rebuild
git clone [email protected]:syntheticChemistry/wetSpring.git
cd wetSpring/barracuda
cargo test --workspace # 1,443+ tests
cargo run --release --bin validate_diversity # Diversity index validation
cargo run --release --bin validate_dada2_full # DADA2 pipeline validation
Key modules: barracuda/src/bio/ (47 CPU + 47 GPU modules), barracuda/src/io/ (FASTQ, mzML, mzXML, JCAMP-DX parsers) Dependency: barracuda (pure math primal, path dependency to ../../barraCuda/crates/barracuda)
2. Pharmacometric Modeling (vs NONMEM / Monolix / WinNonlin)
What It Replaces
| Tool | Annual Cost | What It Does |
|---|
| NONMEM | ~$2,000/yr | Population PK parameter estimation (FOCE) |
| Monolix | ~$1,500/yr | Population PK parameter estimation (SAEM) |
| WinNonlin (Phoenix) | ~$3,000/yr | Non-compartmental analysis (NCA) |
| CRO population PK | $50K–200K/program | Contract research organization modeling |
ecoPrimals Replacement: healthSpring
Repository: github.com/syntheticChemistry/healthSpring Status: V35 — 613 tests, 73 experiments, 113/113 cross-validation checks, 6 WGSL shaders
| Capability | Parity Level | Notes |
|---|
| Hill dose-response (4-parameter) | Full parity | Validated for JAK inhibitors (Gonzales IC50 data) |
| One-compartment PK (IV bolus, oral Bateman) | Full parity | AUC trapezoidal, steady-state accumulation |
| Two-compartment PK (biexponential) | Full parity | Distribution/elimination phase separation |
| mAb PK cross-species (allometric scaling) | Full parity | BW^0.75 CL, BW^1.0 Vd |
| Population PK Monte Carlo (1,000+ patients) | Full parity | Lognormal IIV, CL-AUC correlation |
| PBPK (5-tissue physiological) | Full parity | Mass conservation, hepatic clearance, tissue Kp |
| Michaelis-Menten nonlinear PK | Full parity | Capacity-limited elimination (phenytoin) |
| FOCE estimation | Near parity | 30% theta recovery on synthetic data |
| SAEM estimation | Near parity | 50% theta recovery on synthetic data |
| NCA (λz, AUC∞, MRT, CL, Vss) | Full parity | All 5 standard NCA metrics |
| NLME diagnostics (CWRES, VPC, GOF) | Full parity | CWRES ~N(0,1), 50-simulation VPC |
| GPU population Monte Carlo | Exceeds | 207 M/s throughput (RTX 4070), 100K patients |
| Capability | Gap | Path to Parity |
|---|
| FOCE/SAEM on real clinical data | Validated on synthetic only | Need MIMIC-IV (PhysioNet credentialed access) |
| FDA submission formatting | No CTD/eCTD output | Infrastructure exists; formatting layer needed |
| Covariate model building (stepwise) | Manual covariate selection only | Automated stepwise selection planned |
| Regulatory track record | No FDA submissions yet | Deterministic reproducibility is an advantage for auditors |
| Interactive model exploration | CLI only | 🌸👅 petalTongue dashboard provides visualization |
| Existing training / certification | No formal training program | K-Nome methodology available |
Where to Find / Rebuild
git clone [email protected]:syntheticChemistry/healthSpring.git
cd healthSpring
cargo test --workspace # 613 tests
cargo run --release --bin exp001_hill # Hill dose-response (Gonzales IC50)
cargo run --release --bin exp004_mab # Cross-species PK (lokivetmab → human)
cargo run --release --bin exp075_nlme # NONMEM/Monolix/WinNonlin replacement
Key modules: ecoPrimal/src/pkpd/ (compartmental, population, NLME), ecoPrimal/src/microbiome/ (gut Anderson), ecoPrimal/src/biosignal/ (ECG, PPG, EDA) Rust vs Python: 84× aggregate speedup across 14 benchmark cases (Exp084)
3. Drug Repurposing (vs Every Cure MATRIX / ROBOKOP)
What It Replaces
| Tool | Cost | What It Does |
|---|
| Every Cure MATRIX | $48.3M ARPA-H grant | Drug-disease scoring (4K drugs × 18K diseases) |
| ROBOKOP | NIH-funded | Knowledge graph for drug-disease relationships |
| DrugBank | $20K+/yr (commercial) | Drug target database |
| repoDB | Free | Drug-disease benchmark dataset |
ecoPrimals Replacement: wetSpring Track 3 + neuralSpring nS-06 + groundSpring
| Capability | Parity Level | Notes |
|---|
| Pathway-based drug scoring (PI3K/AKT/mTOR) | Full parity | Fajgenbaum 2019 reproduced (Exp157, 8/8) |
| MATRIX 4-stage pipeline | Full parity | Pharmacophenomics pipeline (Exp158, 9/9) |
| NMF drug-disease factorization | Full parity | Yang 2020 + repoDB (Exp159–160, 16/16) |
| TransE knowledge graph embedding | Full parity | ROBOKOP-style (Exp161, 7/7) |
| Anderson geometry-aware drug scoring | No equivalent | Novel — adds tissue penetration physics to MATRIX |
| Tissue lattice + barrier promotion | No equivalent | 3D Anderson Hamiltonian for skin/gut geometry |
| Cross-species PK translation | Partial parity | Validated canine→human; not yet feline→human |
| GPU compound screening | Exceeds | 207 M/s Hill sweep on RTX 4070 |
What Every Cure / ROBOKOP Still Do Better
| Capability | Gap | Path to Parity |
|---|
| Scale: 4K drugs × 18K diseases | We validate on 6 drugs × 6 diseases | Data pipeline, not algorithm — ChEMBL + NCATS Translator |
| Curated disease pathway profiles | Use published pathways only | NCATS Translator API (same source as Every Cure) |
| Clinical outcome integration | Published parameters only | MIMIC-IV, FAERS via openFDA |
| Institutional backing / regulatory relationships | One developer | MSU Drug Discovery + ADDRC collaboration |
| Pre-built ontology mappings (MONDO, DO) | Manual mappings | MONDO/DO are open; integration work needed |
Where to Find / Rebuild
# wetSpring — Track 3 drug repurposing
cd wetSpring/barracuda
cargo run --release --bin validate_fajgenbaum_pathway # Exp157
cargo run --release --bin validate_matrix_pharmacophenomics # Exp158
cargo run --release --bin validate_nmf_drug_repurposing # Exp159
cargo run --release --bin validate_repodb_nmf # Exp160
cargo run --release --bin validate_knowledge_graph_embedding # Exp161
# neuralSpring — immunological Anderson + MATRIX scoring
cd neuralSpring
cargo run --release --bin validate_immunological_anderson # 20/20
cargo run --release --bin validate_immunological_anderson_extended # 28/28
# groundSpring — tissue Anderson drug scoring
cd groundSpring
cargo run --release --bin validate_tissue_anderson # Exp033–034
4. Analytical Chemistry (vs MassHunter / Chromeleon / MZmine)
What It Replaces
| Tool | Annual Cost | What It Does |
|---|
| MassHunter (Agilent) | ~$10K+/yr | LC-MS data acquisition + analysis |
| Chromeleon (Thermo) | ~$5K+/yr | Chromatography data system |
| MZmine | Free | LC-MS feature extraction |
| FindPFAS | Free | PFAS mass spectrometry screening |
ecoPrimals Replacement: wetSpring Track 2
| Capability | Parity Level | Notes |
|---|
| mzML parsing | Full parity | Sovereign XML parser (no quick-xml) |
| mzXML parsing | Full parity | Sovereign parser |
| JCAMP-DX parsing | Full parity | Spectroscopy format |
| EIC extraction | Full parity | m/z window extraction from raw data |
| Peak detection (signal processing) | Full parity | CWT-based, validated against Python baseline |
| Spectral cosine matching | Exceeds | GPU kernel, 1,077× speedup |
| KMD grouping | Full parity | Kendrick mass defect for homologous series |
| PFAS screening (mass defect + RT) | Full parity | FindPFAS algorithm reproduced |
| Retention index (Kovats, Lee) | Full parity | Both linear and polynomial calibration |
| Capability | Gap | Path to Parity |
|---|
| Instrument control / data acquisition | Software only — no instrument drivers | Not planned; focus is analysis |
| Vendor-specific raw formats (.d, .raw) | mzML/mzXML only (open formats) | Vendor conversion is standard practice |
| Method development wizards | CLI only | Not planned |
| FDA 21 CFR Part 11 compliance | Provenance chain exists; no formal audit | 🐻🐕 BearDog signing + 🪨📖 loamSpine certs map to Part 11 |
| Integrated LIMS | No LIMS — provenance trio provides chain-of-custody | Paper 21 architecture covers this |
Where to Find / Rebuild
cd wetSpring/barracuda
cargo run --release --bin validate_features # EIC + peak detection
cargo run --release --bin validate_peaks # Signal processing
cargo run --release --bin validate_pfas_decision_tree # PFAS screening
cargo run --release --bin validate_massbank_gpu_scale # GPU spectral matching
Key modules: barracuda/src/io/mzml/, barracuda/src/io/mzxml/, barracuda/src/bio/eic.rs, barracuda/src/bio/signal/
5. Biosignal Processing (vs LabChart / MATLAB / Python-MNE)
ecoPrimals Replacement: healthSpring Track 3
| Capability | Parity Level | Notes |
|---|
| Pan-Tompkins QRS detection | Full parity | Validated against MIT-BIH reference |
| HRV metrics (SDNN, RMSSD, pNN50) | Full parity | Time-domain from R-peak intervals |
| PPG SpO2 calibration | Full parity | Beer-Lambert AC/DC ratio |
| EDA tonic/phasic decomposition | Partial | Sovereign implementation; numpy convolution still faster for rolling average |
| Arrhythmia beat classification | Full parity | Template matching: Normal/PVC/PAC/BBB |
| WFDB format parsing | Full parity | PhysioNet Format 212/16 + beat annotations |
| Multi-channel fusion | Full parity | ECG + PPG + EDA → composite health assessment |
6. Provenance & Data Integrity (vs LabArchives / Benchling / LIMS)
ecoPrimals Replacement: SCYBORG Provenance Trio + BearDog
| Capability | Parity Level | Notes |
|---|
| Sample chain-of-custody | Exceeds | Cryptographic DAG ( 🌱🔐 rhizoCrypt), not database records |
| Ed25519 digital signatures on results | No equivalent | 🐻🐕 BearDog signs every computation result |
| ISO 17025/15189 traceability mapping | Architectural parity | 🪨📖 loamSpine certificates map to ISO requirements |
| Fraud detection (6 types) | Exceeds | Graph analysis: phantom sample, broken cold chain, etc. |
| Consent-gated access (medical) | No equivalent | Paper 22: DID-based consent certificates |
| Audit trail | Exceeds | Every DAG vertex is immutable, signed, and attributed |
What LIMS / Benchling Still Do Better
| Capability | Gap | Path to Parity |
|---|
| Inventory management | No physical inventory tracking | fm-pipette (FIELDMOUSE) planned for wet lab integration |
| Barcode / QR scanning | No hardware integration | 🪺🔒 NestGate mobile scanning planned |
| Regulatory pre-validation (GxP) | Architecture exists; no formal GxP audit | 🐻🐕 BearDog + 🪨📖 loamSpine architecture maps to GxP; audit needed |
| SaaS convenience | Local deployment only | ⚛️🧬 NUCLEUS enables LAN/WAN; no cloud planned |
7. GPU Scientific Computing (vs CUDA / Kokkos / MATLAB Parallel)
| Capability | Parity Level | Notes |
|---|
| f64 precision GPU compute | Full parity | All WGSL shaders use f64 (via DF64 emulation where needed) |
| Vendor-agnostic GPU targeting | Exceeds | WebGPU: NVIDIA + AMD + Intel + Apple (CUDA is NVIDIA-only) |
| Sovereign shader compiler | Exceeds | 🪸🌊 coralReef compiles WGSL without vendor toolchain |
| Hardware discovery | Exceeds | toadStool discovers GPU + CPU + NPU at runtime |
| Mixed hardware dispatch | Exceeds | CPU + GPU + NPU routing by capability, not hardcoding |
| Neuromorphic (NPU) support | No equivalent | BrainChip AKD1000 via pure Rust driver |
| 806+ validated WGSL shaders | Specialized | Bio/physics domain; not general-purpose |
What CUDA / Kokkos Still Do Better
| Capability | Gap | Path to Parity |
|---|
| Raw throughput (CUDA optimized kernels) | WGSL overhead for some operations | Improving with wgpu maturity |
| Ecosystem (cuBLAS, cuFFT, cuDNN) | 🐟⚡ barraCuda covers science ops; no ML framework | Not competing with ML frameworks |
| Multi-GPU scaling | Single GPU per dispatch | toadStool multi-device dispatch planned |
| Tensor cores / mixed precision | f64 focus, not fp16/bf16 | Science needs f64; ML-style mixed precision not a priority |
| HPC job scheduler integration (Slurm) | No Slurm scripts | ⚛️🧬 NUCLEUS bonding model is an alternative; Slurm adaptor trivial |
8. Aggregate Ecosystem Metrics
| Metric | Value |
|---|
| Springs (validation domains) | 7 (wet, hot, air, ground, neural, health, ludo) |
| Primals (infrastructure) | 14 ( 🐟⚡ barraCuda, toadStool, 🪸🌊 coralReef, 🌿🖥️ biomeOS, 🐻🐕 BearDog, 🪺🔒 NestGate, 🎵🐦 Songbird, 🍯🌾 sweetGrass, 🌱🔐 rhizoCrypt, 🪨📖 loamSpine, 🌸👅 petalTongue, 🐿️🧠 Squirrel, 🎲🧊 bingoCube, FIELDMOUSE) |
| Total tests | 27,000+ across all springs |
| Total validation checks | 15,334+ |
| Papers reproduced | 70+ (63 💧♨️ wetSpring + ❤️♨️ healthSpring + 🧠♨️ neuralSpring + 🔥♨️ hotSpring + others) |
| WGSL shaders ( 🐟⚡ barraCuda) | 806+ |
| Languages | Pure Rust (zero C/C++/Fortran in application code) |
| Unsafe code | Zero (#![forbid(unsafe_code)] in all spring lib crates) |
| External dependencies | Minimal; all pure Rust or explicit rust_backend |
| License | AGPL-3.0-or-later (code), CC-BY-SA-4.0 (docs) |
| Development methodology | K-Nome (Knowledge-Numeric Observed & Mentored Evolutionary Programming) |
| Development history | 69,000+ AI invocations, 51B tokens, 185-day streak |
| Hardware investment | ~$15,000 (consumer hardware, zero cloud) |
These capabilities have no commercial equivalent:
| Capability | What It Does | Spring |
|---|
| Anderson localization for community structure | Maps microbial diversity onto condensed matter physics; predicts signal propagation vs confinement | 💧♨️ wetSpring, ⛰️♨️ groundSpring, 🧠♨️ neuralSpring |
| Geometry-aware drug repurposing | Adds spatial tissue penetration to pathway-based drug scoring (extends MATRIX) | 💧♨️ wetSpring Track 3, 🧠♨️ neuralSpring nS-06, ⛰️♨️ groundSpring |
| Cross-species Anderson translation | Same physics, different tissue parameters — species-agnostic by construction | ❤️♨️ healthSpring Track 6 |
| Cryptographically signed scientific results | Ed25519 signature on every diversity index, ASV table, drug score | 🐻🐕 BearDog |
| Provenance DAG for biological samples | Field-to-publication chain-of-custody with fraud detection | 🌱🔐 rhizoCrypt + 🪨📖 loamSpine + 🍯🌾 sweetGrass |
| NPU edge classification | BrainChip AKD1000 at 18.8K Hz, coin-cell power, pure Rust driver | toadStool + 🧠♨️ neuralSpring |
| Sovereign shader compiler | Compile WGSL without NVIDIA/AMD/Intel toolchain | 🪸🌊 coralReef |
| Zero-knowledge medical provenance | Patient-owned records with consent certificates | Paper 22 |
10. Rebuilding From Source
Every spring is a self-contained Cargo workspace. To rebuild any capability:
# Prerequisites: Rust 1.87+ (rustup.rs), git
# Optional: GPU (any Vulkan-capable card for GPU tests)
# Clone the spring you need
git clone [email protected]:syntheticChemistry/<spring>.git
cd <spring>
# Build and test
cargo test --workspace # All library tests
cargo clippy --all-targets # Zero warnings guaranteed
cargo run --release --bin <binary> # Run specific validation
# GPU tests (requires Vulkan-capable GPU)
cargo test --workspace --features gpu
Build dependencies: Rust toolchain only. No Python, no R, no conda, no Docker, no pip, no npm. One cargo build compiles everything from source.
🐟⚡ barraCuda (the math primal) is a path dependency. Clone it alongside the spring:
cd Development/ecoPrimals
git clone [email protected]:ecoPrimals/barraCuda.git # renamed from barraCUDA
git clone [email protected]:syntheticChemistry/wetSpring.git
# Cargo.toml points to ../../barraCuda/crates/barracuda
Document History