ecoPrimals for Principal Investigators — What This Actually Replaces in Your Lab

From: ecoPrimal — human + synthetic intelligence
Organization: 🔧🦎 ecoPrimals Date: March 17, 2026 Repositories: github.com/ecoPrimals — all AGPL-3.0-or-later

The Short Version

Your lab probably runs some combination of Python/R bioinformatics, commercial pharmacometric software, and ad hoc data management. 🔧🦎 ecoPrimals is a pure Rust stack that replaces most of that with faster, reproducible, GPU-accelerated alternatives — and adds things no commercial tool does (cryptographic provenance, physics-based drug scoring, vendor-agnostic GPU compute).

Every claim below has a cargo run --bin validate_* binary that proves it. You can clone the repo and verify on your own hardware.

Find Your Domain

Physics & Materials — The 🪨✅ guideStone deployment artifact validates published lattice QCD, plasma physics, molecular dynamics, and spectral theory results on commodity hardware. A single binary, no CUDA, no vendor SDK. Consumer GPUs do real f64 science via Vulkan. See the guideStone section and Papers 01, 06, 07, 10, 14, 23, 25.

Pharmacology & Immunology — The ⛺📄 baseCamp paper program has reproduced dose-response curves, pharmacokinetics, tissue-geometry modeling, and drug repurposing scoring from published veterinary and human data. Anderson localization applied to cytokine signaling is original work. See Papers 12, 13, 22.

Microbiology & Genomics — Sovereign 16S pipelines, metagenomics, phylogenetics, PFAS detection, and quorum sensing models — all in pure Rust, all reproducing published results. See Papers 02, 03, 04, 05, 09, 16.

Game Science & Creative Computing — Rigorous HCI models, game design as science, distributed computation, and 🔮🕸️ esotericWebb as a proof that sovereign infrastructure produces real creative software. See Papers 17, 18, 19, 24.

What You Actually Save

Minimum annual savings: $6,500 in licenses alone. CRO avoidance: $50K+ per program.

What You Actually Get That’s Better

Speed

Reproducibility

• Every validation binary has hardcoded expected values with explicit tolerances
• #![forbid(unsafe_code)] — zero undefined behavior, guaranteed by the compiler
• cargo clippy with pedantic + nursery lints: zero warnings across every spring
• Deterministic: same input → same output, always. No Jupyter state, no Python version drift
• One build command: cargo build --release. No conda, no pip, no Docker, no sysadmin

What No Commercial Tool Offers

How It Works With Your Existing Infrastructure

If You Have Sequencing (Genomics Core / RTSF)

Your sequencer → FASTQ files
    → wetSpring 16S pipeline (FASTQ→QC→merge→derep→DADA2→chimera→taxonomy→diversity→UniFrac)
    → Anderson localization analysis (novel community structure physics)
    → Provenance chain (every step signed, auditable, ISO-mappable)

💧♨️ wetSpring replaces the Galaxy/QIIME2/mothur/R pipeline with a single cargo run. All 306 validation binaries pass. 63 published papers reproduced.

If You Have an HTS Core (like ADDRC)

Compound library → plate reader → IC50 data
    → healthSpring GPU Hill sweep (207 M/s on RTX 4070)
    → MATRIX pathway scoring (Fajgenbaum 2019 reproduced)
    → Anderson geometry scoring (tissue penetration physics)
    → Ranked candidates → back to wet lab validation

The GPU shader can score 8,000 compounds × 6 cytokine pathways in seconds. Traditional screening informatics (ActivityBase, GREENScreen) scores compounds but doesn’t consider tissue geometry — drugs that can’t reach their target score well in silico but fail in vivo.

If You Have ICER Access

ICER A100 allocation → barraCuda WGSL shaders (vendor-agnostic)
    → Anderson eigensolve at L=200 (production scale)
    → Population PK at 10M patients
    → MATRIX scoring at 4K × 18K scale (72M evaluations)

WGSL shaders compiled by wgpu run on any Vulkan-capable GPU. No CUDA required. No NVIDIA lock-in. The same binary that runs on your lab’s RTX 3060 runs on ICER’s A100s.

If You Run Clinical Trials

Patient data → healthSpring PK/PD pipeline
    → NONMEM-equivalent FOCE estimation (sovereign, no Fortran)
    → NCA (λz, AUC∞, MRT, CL, Vss — WinNonlin replacement)
    → NLME diagnostics (CWRES, VPC, GOF)
    → petalTongue visualization → clinical dashboard
    → BearDog signed results → audit trail

Every intermediate result is signed. The provenance chain maps to 21 CFR Part 11 requirements. No Fortran compiler. No proprietary binary.

What We Honestly Can’t Do Yet

How To Evaluate

# Pick the spring relevant to your domain:
git clone [email protected]:syntheticChemistry/healthSpring.git  # PK/PD, clinical
git clone [email protected]:syntheticChemistry/wetSpring.git     # 16S, metagenomics, LC-MS

# Build (requires Rust 1.87+ from rustup.rs — 2 minute install)
cd healthSpring && cargo test --workspace   # 613 tests, 0 failures
cd wetSpring/barracuda && cargo test --workspace  # 1,443 tests, 0 failures

# Run a specific validation
cargo run --release --bin exp001_hill       # Hill dose-response
cargo run --release --bin validate_diversity # Shannon/Simpson/Pielou/Chao1

# No Python. No R. No conda. No Docker. No licenses. No cloud account.

Published Work Reproduced With Full Provenance

The springs reproduce published, peer-reviewed science as acceptance tests for the infrastructure. Each entry below is a researcher whose published work has been independently reimplemented in Rust, cross-validated against the original Python/R/MATLAB results, and promoted to GPU — with every check automated, every tolerance explicit, and every result fully public under the ⚖️🔓 scyBorg license.

Total across all springs: 70+ papers reproduced, 107,000+ test functions, 614K lines of Rust. Every reproduction is executable: cargo run --release --bin validate_* reproduces the result on your hardware.

Contact

ecoPrimal — github.com/ecoPrimals Written and developed by ecoPrimal: human + synthetic intelligence. Built on ~$15,000 of consumer hardware. Zero cloud bills. Zero licenses.