Skip to content
OpenAlgo
NewWet-Lab Launcher: plan a lab at your ZIP code

Discovery shouldn’t need an institution.

OpenAlgo turns published chemistry into runnable, audited code in minutes, and turns a lab idea into a costed, permitted plan for your address. The leverage of a research department, without the gatekeepers.

Start with a paper

Private beta: approved accounts translate today · Request access

Starting a lab instead? Plan it with the Wet-Lab Launcher

Fig. 1 — One paper, translated

  1. extract
  2. review
  3. generate
Illustrative. Real sessions link every field to the sentence it came from.Templates, not freeform generation

The bottleneck

Science has a throughput problem. It isn’t a shortage of ideas.

More than

70%

of researchers have tried and failed to reproduce another scientist’s experiments.

Baker, M. Nature 533, 452–454 (2016). Survey of 1,576 researchers. Source

01

Methods are trapped in PDFs

The recipe is published; the code usually isn’t. Every team re-implements the same pipeline from prose, one ambiguous sentence at a time.

02

Weak papers cost weeks

You learn that a result doesn’t hold only after you’ve built it. That is weeks of a senior scientist’s time spent proving a dead end.

03

Benches are gated

An idea with no institution behind it stalls at the lab door: capital, equipment and permits that nobody lays out up front.

Two products · one conviction

From paper to pipette.

The mechanical work of science should be automated, audited and open to anyone who can connect the dots. We start at both ends: the literature, and the lab bench.

OpenAlgo ResearchPrivate beta

Published paper → runnable, audited code

Know in minutes whether a paper holds up.

  • DOI or PDF in, reviewed Python project out
  • Every value traced to the sentence it came from
  • Every missing detail reported in REPRODUCIBILITY_GAPS.md
Explore Research Translate a paper

Wet-Lab LauncherNew

Lab idea → costed, permitted plan

Cost, equip and permit a lab at your address.

  • Launch cash, deposits and monthly run-rate
  • Which instruments to own and which to share
  • Permits routed to your state, county and city
Explore the Launcher Start a plan, free

How they connect

Read it. Run it. Then test it at your own bench.

Most tools stop at the PDF or start at the bench. The loop only closes when both are covered.

  1. 01

    Read

    A promising method lands in the literature.

  2. 02

    Reproduce

    Methods become a runnable project, gaps flagged.

  3. 03

    Decide

    Kill weak papers in minutes, not after a month of coding.

  4. 04

    Launch

    Cost, equip and permit a bench at your address.

  5. 05

    Measure

    Your own first reproducible result.

OpenAlgo Research →covers 01–03Wet-Lab Launcher →plans 04–05

OpenAlgo Research

An 80% codebase in minutes. The other 20% is science.

QSAR pipelines share one architecture: molecules in, featurize, split, model, evaluate. That standardization is what makes paper-to-code a solvable problem instead of a demo.

How Research works
Input01

SMILES + targets

Dataset, source, cleaning

Featurize02

Fingerprints · descriptors · graphs

Morgan radius, bits, RDKit set

Split03

Random · scaffold · temporal

Ratios, stratification, seed

Model04

RF · XGBoost · SVM · GNN

Architecture, hyperparameters

Evaluate05

RMSE · R² · ROC-AUC

Matched to what the paper reports

Provenance

field: split_strategy

Extracted value

Scaffold split · 80 / 10 / 10

“…molecules were partitioned with a Bemis–Murcko scaffold split (80/10/10) to limit leakage between train and test…”Section 2.3 · page 4

Read A

✓ scaffold 80/10/10

Read B

✓ scaffold 80/10/10

Confidence: high — both reads agree, quoted source found

100

papers with public gap reports in the Reproducibility Hub

5

template families, from fingerprint models to graph neural networks

2

independent reads of every paper, with disagreements surfaced to you

Wet-Lab Launcher · New

Your own bench, without an institution behind you.

Enter a ZIP code and a first workflow. Get the cash to launch and run, the instruments to own versus share, the permits that apply at your address, and the order to do it in. Four steps. We email you the plan.

U.S. jurisdictions: all 50 states and DC
51U.S. jurisdictions: all 50 states and DC
international regimes: England, Germany, Singapore, India
4international regimes: England, Germany, Singapore, India
local permits routed to the right city or county
ZIPlocal permits routed to the right city or county
a decision for every major instrument
Own · sharea decision for every major instrument
Start your plan, free See what a plan covers

Planning estimates, not quotations or legal advice.

Our standard

We’d rather tell you a paper won’t reproduce in five minutes than let you find out in five weeks.

01

Every assumption surfaced

If the paper didn’t say it, we don’t pretend it did. Gaps are flagged and every default is named.

02

Templates, not improvisation

Code comes from tested templates, so the same reviewed paper gives you the same project every time.

03

You sign off

Nothing reaches code without your review. The scientist stays the scientist.

04

Honest scope

QSAR and molecular property papers; biochemical and BSL-1 labs. We say plainly what we don’t cover yet.

Who moves first

For scientists who don’t wait for permission.

Independent scientists and founders

Test a method in silico, then plan the bench to prove it, without waiting on an institution.

Plan a lab

Computational chemists

Get the scaffold in minutes and spend your judgment on the part that needs a scientist.

Translate a paper

Cheminformatics engineers

Clean, pinned project structure with proper RDKit setup, not a notebook to reverse-engineer.

See coverage

Discovery leads

A fast-fail filter for the literature, with private deployment and your own templates.

Enterprise

Questions

Asked by skeptical scientists

General-purpose models write code from scratch every time: different structure, different bugs, no chemistry-specific defaults. OpenAlgo uses AI only to extract parameters, makes you verify them against the source text, then fills tested templates with built-in RDKit configuration, salt stripping, scaffold splits and descriptor scaling. The output is consistent and auditable.

Every paper is read twice with different strategies. Where the reads disagree you see both values side by side and choose. Nothing reaches generated code without your sign-off.

QSAR and molecular property prediction: classification and regression with descriptors, fingerprints or graph representations. Docking, molecular dynamics and retrosynthesis are out of scope for now. The coverage page lists each template family and its known limits.

Planning estimates in USD for a two-person team doing biochemical or BSL-1 molecular work. A ZIP code gives you state rules in all 50 states and DC, local permits routed to the right city or county, and metro cost indices; curated city rules cover a handful of places so far. England, Germany, Singapore and India are available as international regimes. Estimates are not quotations or legal advice.

On the shared cloud, uploaded papers may be stored. Teams that cannot send proprietary PDFs to shared infrastructure can plan a customer-owned deployment with Enterprise. See the privacy policy for details.

Research is in private beta: approved accounts start on the free tier, downloads start with the Researcher plan at $49 a month, and teams get a custom pilot. The Wet-Lab Launcher is free. Details on the pricing page.

The literature won’t reproduce itself.

Every week spent re-implementing a paper by hand is a week not spent on your own molecules. Start with one DOI, or one ZIP code.

Translate a paper Plan a wet lab