Chemical perturbations,
measured at depth.

VCPI profiles compounds by DRUG-seq in human THP-1 cells and releases every sample openly. Submit your compounds for free. Download everything that has been released.

Browse datasetsSubmit compounds
Samples
97,986
Compounds
26,236
Releases
4
Cells
THP-1, 24 h
123456789101112131415161718192021222324ABCDEFGHIJKLMNOP
[VCPI-0001] one of 77 plates30 nM to 10 µMDMSOReference control
Run onGinkgo Datapoints

Released so far

Two dose-response releases for potency, two single-dose releases for breadth. All free to registered users.

Open the dataset browser
AccessionDesignCompoundsSamplesReleased
[VCPI-0004]ReFRAMESingle dose12,21027,7862026-10-23
[VCPI-0003]Single dose10,26121,8882026-05-20
[VCPI-0002]Dose-response1,49819,1402026-04-01
[VCPI-0001]Dose-response2,28229,1722026-03-18

[VCPI-0004]

The ReFRAME library, profiled

ReFRAME is a collection of more than 12,000 compounds that have reached clinical development or undergone significant preclinical profiling, assembled by the Calibr-Skaggs Institute for Innovative Medicines at Scripps Research from commercial databases and patent filings. Many were synthesized because no supplier carried them.

VCPI-0004 measures ReFRAME compounds at 10 µM in two replicates after 24 hours in THP-1 cells, with transcriptional and cytotoxicity controls on every plate. The screen covers more than 12,000 ReFRAME compounds across VCPI-0004 and earlier releases. A six-dose follow-up on the top 10% of hits is planned. Phenotypic screening results for the same molecules are searchable at reframedb.org.

ReFRAME screened
12,949
In VCPI-0004
12,210
Dose
10 µM
License
CC BY 4.0
Read the release noteJanes et al., PNAS 2018

Library assembled by

Calibr-Skaggs Institute for Innovative Medicines, Scripps Research

From your compound to an open release

  1. 1

    Submit

    Upload a CSV of SMILES from the Submit Compounds page. Screening is free, and a paid embargo is available on request.

  2. 2

    Screen

    Ginkgo Datapoints runs DRUG-seq on THP-1 cells after 24 hours of treatment. Dose-response releases test six doses from 30 nM to 10 µM; screening releases test one dose across many more compounds.

  3. 3

    Release

    Every sample is published to registered users: download from the browser, query with the vcpi-client Python package, or export a custom slice across releases.

Work in Python

The vcpi client runs SQL over sample metadata and chemistry without downloading anything, then pulls full count matrices when you need them. Create an access token on the Settings page.

virtualcell-vcpi/vcpi-client
# pip install git+https://github.com/virtualcell-vcpi/vcpi-client.git
import vcpi

vcpi.list_datasets()

# SQL over metadata and chemistry, nothing downloaded
df = vcpi.query(
    job="vcpi-0004",
    sql="SELECT * FROM metadata WHERE percent_mitochondrial < 20",
)

# Full count matrix, metadata and chemistry for one release
exp = vcpi.load_experiment("vcpi-0001")
exp["data"]  # genes x samples

Built on the releases

Partners add analyses and model outputs next to the measured data.

  • Ginkgo Datapoints

    Runs the assays and sequencing on its automated foundry and hosts the open releases.

  • QIAGEN

    1,452 IPA pathway analyses, one per compound and dose, linked from three releases.

  • Synthesize Bio

    AI-generated virtual replicates published next to the wet-lab files for three releases.

  • DeepLife

    TwinCell pilot: causal target predictions for six compounds across eight targets.

Why DRUG-seq

Depth per condition
DRUG-seq reads ~10,000 genes per condition in arrayed wells. Pooled single-cell methods read ~1,500 genes per cell.
Dose-response built in
Compounds in dose-response releases are profiled at six concentrations with replicates, so potency is part of the data.
One cell type, controlled
Every sample is THP-1 at 24 hours, with DMSO and reference controls on every plate. No cross-lab batch guesswork.
Ginkgo’s DRUG-seq offering

Get access

Create a free account to download every release and submit compounds for screening.

or

By signing up you agree to the terms of our privacy policy

Questions

What is VCPI?

The Virtual Cell Pharmacology Initiative profiles compounds by DRUG-seq in a single human cell type and releases the results openly, so researchers can study chemical perturbations and train models on consistent data.

Is this a prediction tool or a wet lab?

A wet lab. Every sample is a real well: real compounds, real cells, real sequencing. Partners build analyses and models on top of the releases.

Who can access the data?

Any registered user. Sign up, then download any release from the browser or with the Python client. Academic, industry and independent researchers are all welcome.

How do I contribute compounds?

Sign in, open Submit Compounds and upload a CSV with SMILES. Screening is free. Compounds join a future release once sequencing and quality control are complete.

Can I protect sensitive molecules?

Yes. A paid embargo is available on request: mark the embargo option on your submission and the team will follow up. No chemistry needs to be disclosed to start that conversation.

Which cell line and conditions do you use?

THP-1, a human monocytic leukemia line, treated for 24 hours. Plates carry DMSO vehicle and reference control wells.

Is a Virtual Cell model available today?

Not from VCPI itself. The releases are the training and benchmarking foundation; partner analyses such as QIAGEN IPA and DeepLife TwinCell are available on the Analyses page.