Gruffi: an algorithm for computational removal of stressed cells from brain organoid transcriptomic datasets

Area of Science:

• Developmental Biology
• Computational Biology
• Genomics

Background:

• Organoids are valuable in vitro models for studying development and disease.
• 3D organoid cultures, particularly brain organoids, suffer from insufficient oxygen, leading to cellular stress.
• This stress can negatively impact lineage commitment and the accuracy of developmental trajectory analysis.

Purpose of the Study:

• To identify and characterize cellular stress in organoid models.
• To develop a computational method to remove stressed cells from organoid single-cell RNA sequencing (scRNAseq) data.
• To improve the accuracy of organoid models for studying developmental processes.

Main Methods:

• Analysis of approximately 190,000 single-cell RNA sequencing (scRNAseq) cells from brain organoids and fetal samples.
• Identification of a unique stress signature present in organoid data but absent in fetal data.
• Development and validation of a computational algorithm, Gruffi, for unbiased identification and removal of stressed cells.

Main Results:

• A distinct subpopulation of stressed cells was identified in all organoid samples, not present in fetal samples.
• Cellular stress was found to be confined to this subpopulation and did not affect neuronal specification or maturation.
• The Gruffi algorithm successfully identified and removed stressed cells across various organoid datasets and protocols.
• Validation extended to retinal organoids, demonstrating broad applicability.

Conclusions:

• Cellular stress is a common artifact in organoid cultures that can be computationally corrected.
• The Gruffi algorithm provides an unbiased method to improve the analysis of organoid scRNAseq data.
• Bioinformatic correction of cell stress enhances the delineation of developmental trajectories and the resemblance of organoid data to in vivo findings.