Quick Start

This walkthrough uses the built-in test profile to demonstrate the complete two-phase workflow. We suggest you start by running the local CLI example. If you will be using Seqera Platform for production runs, proceed to running the test profile there after running the local CLI example.

Step 1: Run the Test Profile (Phase 1)¶

The test profile automatically downloads a small dummy dataset containing:

• A minimal subset of images from one well (both Cell Painting and Barcoding arms)

• Pre-configured CellProfiler pipelines

• A sample Barcodes.csv file

The pipeline will:

1. Download the test dataset

2. Run illumination calculation and application

3. Run segmentation (Cell Painting) and barcode calling (Barcoding)

4. Generate QC montages

5. Stop before stitching (this is expected behavior)

Local CLI¶

nextflow run broadinstitute/nf-pooled-cellpainting \
    -profile test,docker \
    --outdir results

Expected runtime: 15-20 minutes locally

Seqera Platform¶

In Seqera Platform, on the sidebar select Launchpad and then click Add Pipeline.

Set the following variables:

Select Add

Next to the newly created test-pipeline from the Launchpad select Launch.

In the Run parameters of the test-pipeline, enter an S3 path where your data should be output into outdir (e.g. s3://my-bucket/output-folder/) and select Launch.

In the Runs view you can see your newly created run. If you click on the run it will show you the status of the workflow modules as they run and complete.

Expected runtime: 5-10 minutes on AWS Batch

Step 2: Inspect QC Outputs¶

On real data, you would check for the following QC criteria. Note that the test data is too small to produce meaningful QC, but you can still look at the outputs to see that they are created:

• Illumination montages: Smooth, gradual intensity variations (not patchy)

• Segmentation previews: Cell/nucleus outlines accurately trace boundaries

• Alignment reports: Small, consistent pixel shifts across the field

Local CLI¶

Navigate to results/workspace/qc_reports/ and inspect the contents of the folders:

results/workspace/qc_reports/
├── 1_illumination_painting/
├── 3_segmentation/
├── 5_illumination_barcoding/
├── 6_alignment/
└── 7_preprocessing/

Seqera Platform¶

Select the Reports tab and select each of the individual reports to examine them.

Step 3: Complete the Run (Phase 2)¶

Resume with QC flags set to true.

Local CLI¶

nextflow run broadinstitute/nf-pooled-cellpainting \
    -profile test,docker \
    --outdir results \
    --qc_painting_passed true \
    --qc_barcoding_passed true \
    -resume

Seqera Platform¶

Open the hamburger menu in the upper right and select “Resume”.

In the run parameters, toggle the qc_painting_passed and qc_barcoding_passed on. Select Launch.

Step 4: Explore Final Outputs¶

After completion, your results/ directory (either local for a local CLI run or in your S3 bucket for Seqera Platform) contains:

results/
├── images/                    # All processed images by batch/stage
│   └── Batch1/
│       ├── illum/
│       ├── images_corrected/
│       ├── images_corrected_cropped/
│       └── images_corrected_stitched/
├── workspace/
│   ├── analysis/              # Final CSV results (most important!)
│   ├── load_data_csv/         # CellProfiler input files
│   └── qc_reports/            # QC visualizations
├── multiqc/                   # Summary report
└── pipeline_info/             # Execution logs and metrics

The most important outputs are in results/workspace/analysis/—CSV files containing linked phenotype and genotype data for every cell.