OPEN IN-SILICO SCIENCE ยท NON-PROFIT

Open in‑silico science, freely shared.

Yepgent Science builds and freely shares open predictive models, harmonized datasets, and honest research — so scientists can work computationally: faster, cheaper, and more reproducible. Non-profit, and open to everyone.

127k+
curated compounds
2
open datasets
1
tool live
$0
open ยท non-profit

The catalog

Open tools, datasets, and research from Yepgent Science — what you can use today and what we’re building next. Each ships with its models, data, and honest, error-barred validation.

Anima

The in-silico safety screen. Enter one molecule → instant cardiotoxicity (hERG) and liver-injury (DILI) triage in a single call.

  • Calibrated risk — a probability and a band, not a bare yes/no.
  • Applicability-domain flag — says off-domain, low confidence when the chemistry is unlike training.
  • Nearest known compounds — the rationale behind every call.
Open the tool →Live

CiPA cardiac panel

Multi-channel proarrhythmia — hERG, Nav1.5, Cav1.2 — toward the FDA CiPA in-silico framework.

Why it matters: one channel can miss real risk; the CiPA panel models the fuller proarrhythmia picture regulators are moving toward.

next endpointIn progress

Open Tox Atlas

Harmonized, provenance-tracked open datasets and model cards across safety endpoints — free to reuse.

Why it matters: harmonized, source-agreeing data — not bigger models — is the proven lever for better safety prediction.

CC-BY-SAPlanned

Research & write-ups

We publish what we learn — methods, dataset cards, and honest findings, including the negative results (like liver injury’s structure-only ceiling). Open, versioned, and reusable, so the field can build on it rather than repeat it.

open · as we goOngoing

Science you can check, run, and shape.

Yepgent Science exists to make rigorous in-silico science open and usable — not locked behind a lab or a paywall. Inspect the methods, run the models, and help decide what gets built next.

Check it

Every model, dataset, and result is openly licensed, versioned, and scaffold-split with error bars. Reproduce the numbers, audit the methods, reuse the data.

Run it

Create an account and run the models directly — each prediction comes with an applicability-domain flag that tells you when to trust the answer, and when not to.

Shape it

Request endpoints, contribute matched data, and review results in the open. The roadmap follows what the community actually needs.

Built in the open — with you.

Yepgent Science is meant to be a shared workspace, not a walled garden. An account lets you run the tools and help steer the work — and everything you contribute stays openly licensed.

Request

In development

Ask for a new endpoint, model, or dataset — and upvote what others have asked for. The roadmap follows the requests.

Contribute

Planned

Have matched assay or exposure data? Submit it through the open schema and get credited in the dataset card.

Discuss

Planned

Comment on models and results, post what you find, and review methods in the open. Science gets better under scrutiny.

Create an account Sign in One account across all of Yepgent · collaboration tools are in development.