The landscape of circRNA expression in the human brain

Gokool et al. Biological Psychiatry (2019).


Circular RNAs (circRNAs) are enriched in the mammalian brain and are upregulated in response to neuronal differentiation and depolarisation. These RNA molecules, formed by non-canonical back-splicing, have both regulatory and translational potential. Here, we carried out an extensive characterisation of circRNA expression in the human brain, in nearly two hundred human brain samples, from both healthy individuals and autism cases.

We identify hundreds of novel circRNAs and demonstrate that circRNAs are not expressed stochastically, but rather as major isoforms. We characterise inter-individual variability of circRNA expression in the human brain and show that inter-individual variability is less pronounced than variability between cerebral cortex and cerebellum. Finally, we identify a circRNA co-expression module upregulated in autism samples, thereby adding another layer of complexity to the transcriptome changes observed in autism brain.

These data provide a comprehensive catalogue of circRNAs as well as a deeper insight into their expression in the human brain, and are available as a free resource in browsable format.


CircRNA expression data is available in a browsable format in the Zenbu genome browser.



  • circRNAs from the three datasets DS1, DS2 and organoid cultures are displayed as separate tracks.

  • Each circRNA is displayed as a horisontal bar conecting the ends of  the back splice junction.

  • Color-coding reflects the mean circRNA expression value across all samples (red: high expression, blue: low expression; Click on the "configure track" button to display the color legend ). Note that the color coding adaptively changes for each genomic window displayed.

  • Click on a circRNAs to display expression values in individual samples as a pop-up box.

  • General user instrcutions for Zenbu are avalable here.

  • Note that circRNAs in DS1 and DS2 were filtered for those present in a minimum of 5 samples. See the Methods section of Gokool et al. 2019 for full details of circRNA detection and filtering.