Quickstart

Get ready

A Discovery Loop account is required. Create one free if needed. Lab Explorer is free and needs no payment details.

Have these ready:

• A compatible instrument, or its sample output file (CSV, HPLC export, plate reader file)
• Python 3.9 or newer installed where the adapter will run
• An API key from your Discovery Loop dashboard

Step 1: Add the adapter

Add the Discovery Loop adapter package. It reads instrument output, parses it, and sends results to the loop.

pip install discoveryloop-adapter

Check the install:

dlp --version

Step 2: Sign in

From Settings / API keys, copy your API key, then run:

dlp auth login --key YOUR_API_KEY

The key is stored in ~/.dlp/credentials, so later commands do not need it again.

Step 3: Set up an instrument adapter

Open the adapter wizard:

dlp adapter init

It will prompt you for:

• Instrument type: select one: HPLC, plate-reader, spectroscopy, CSV
• Watch path: the folder where the instrument writes output
• Output signal: the output-file column holding the measurement of interest

The wizard writes ./dlp-adapter.yaml. Edit the file afterward to change its settings.

adapter:
  type: hplc-agilent
  watch_path: /data/hplc/output
  signal_column: area_percent
  run_id_pattern: "{date}_{sample_id}"

Step 4: Set up a research project

Define the result your project will optimize:

dlp project create \
  --name "Reaction yield optimization" \
  --objective maximize \
  --signal area_percent \
  --target 90

Use the --target flag to define when to stop. Once the loop reaches that target, it completes the project. Leave the flag out to continue exploring until you stop it yourself.

Step 5: Set the parameter space

List what the loop may vary:

dlp param add --project my-project --name pH --min 5.5 --max 7.5 --step 0.2
dlp param add --project my-project --name temp_C --min 30 --max 45 --step 1
dlp param add --project my-project --name catalyst --values A,B,C

Step 6: Run the adapter

Launch the adapter. It monitors for new files and sends them to the loop automatically:

dlp adapter start --project my-project

By default, the adapter stays in the foreground. Add --daemon to run it in the background.

Step 7: Queue first experiment

With no data from prior cycles, the loop selects an initial point in the parameter space using a space-filling strategy. Request that first recommendation:

dlp cycle next --project my-project

The command displays recommended conditions. Run them on the instrument. Once the output file lands in the watch path, the adapter collects it and the second cycle begins automatically.

What comes after

Following each run, the loop revises the hypothesis tree, chooses and queues another experiment. Track cycle progress in the dashboard or query it through the CLI:

dlp cycle list --project my-project --tail 5

Once the loop converges or you stop it, the complete cycle log is available as a JSON export. For programmatic access, see the API Reference.

Need help?

For setup problems, contact the team. They can answer implementation questions.