Data Set: Parse 10 Million Human PBMCs in a Single Experiment
/in Parse/by Harshita SharmaUnlock with quick sign up!
Key Takeaways
Analyze 10 million cells across 1,152 samples in a single experiment
Increase statistical power by profiling more cells per sample
Capture detailed cellular responses to perturbations and drug treatments
Figure 1: Experimental Design Overview
Approximately 10 million PBMCs from 12 healthy donors were treated with 90 different cytokines in a single GigaLab experiment, covering 1,092 experimental conditions.
Cells were thawed, washed, and seeded at 1 million cells per well across 12 plates. After 24-hour cytokine treatment, cells were fixed, barcoded, and processed for whole transcriptome sequencing. Libraries were sequenced on the Ultima Genomics platform, achieving ~31,000 reads per cell, with 62.45% cell retention after barcoding.
After data processing with the Parse Analysis Pipeline v1.4.0, integration, and classification, 9,697,974 cells across 18 immune cell types were identified—including rare populations that are typically missed in smaller experiments. Each condition yielded a median of 7,400 cells, enabling high-resolution analysis of immune responses.
Differential expression analysis identified how cytokines influenced gene activity across cell types. Many cytokines triggered strong transcriptional responses, with over 50 genes upregulated per treatment.
Figure 2: Single-Cell UMAP Overview
9,697,974 PBMCs from 12 donors were integrated with Harmony, clustered using Scanpy, and manually annotated, revealing 18 immune cell types present across all donors and experimental conditions.
Figure 3: Cytokine-Induced Gene Changes
A heatmap summarizes the averaged number of genes significantly upregulated (log fold change >0.3, p <0.001) for each cell type and cytokine, highlighting which immune cells respond most strongly to specific cytokine treatments.
Parse 10M PBMC Cytokines Clustering Tutorial
Joey Pangallo, Efi Papalexi – Parse Biosciences, Seattle, WA
Step-by-step example of analyzing 10 million PBMCs treated with cytokines using the Evercode workflow. Covers data loading, preprocessing, Leiden clustering, and generating UMAP plots with Scanpy.
Parse 10M PBMC Cytokines Clustering Tutorial (Downsampled)
Joey Pangallo, Efi Papalexi – Parse Biosciences, Seattle, WA
Same workflow as above, starting with a downsampled dataset of 1 million cells. Ideal for quicker exploration or limited CPU memory setups.
scCODA Parse 10M PBMC Cytokines
Artur Szałata, Dominik Klein, Soeren Becker, Fabian Theis – Helmholtz-Munich
Demonstrates analysis of cell proportion changes across 10 million PBMCs. Shows how using the full dataset improves statistical significance of perturbation effects. Based on scCODA, a Bayesian model for compositional single-cell data analysis (Nat Commun 12, 6876, 2021).
Parse 10M PBMC Cytokines Dask Workflow
Artur Szałata, Dominik Klein, Soeren Becker, Fabian Theis – Helmholtz-Munich
Walks through preprocessing the 10M cell dataset using Dask. Loads data chunk-wise to reduce memory use and demonstrates highly variable gene selection for downstream analysis.
Dataset License: CC BY-NC 4.0 (non-commercial use). Commercial licensing inquiries: support@parsebiosciences.com

Dr. Ebru Boslem
ANZ Market Manager - Research Genomics