Genes were only included in the database when reported in at least half of the studies.

The DEGs have been pre-filtered to speed up the tables on the website, a full list of DEGs analysis can be downloaded in the download section.

For computing the integrated embedding we were using scVI with the batch information as batch key. The DEGs have been computed using a pseudo-bulk approach with edgeR. Whenever the samples where collected from more than a single study we included the study id as latent variable into the design matrix.

Please cite our manuscript under:

Flotho, M., Amand, J., Hirsch, P., Grandke, F., Wyss-Coray, T., Keller, A., and Kern, F. "ZEBRA: A Hierarchically Integrated Gene Expression Atlas of the Murine and Human Brain at Single-Cell Resolution." Nucleic Acids Research, November 6, 2023, gkad990. https://doi.org/10.1093/nar/gkad990.

Please click on Human combined or Mouse combined on the top left to compare DEGs between different brain regions. By selecting one of the four provided datasets (Human and mouse (non)-cortex) you can get back to the original views.

You can find an overview of the integrated datasets under information. The link column links to the original datasets. Some datasets are not publicly available and can only be accessed after registration or by request to the authors. Also, the download formats may be different and require additional processing to load them into Seurat or Scanpy.

The isoform matrix was derived entirely on the raw data without any preprocessing or filtering steps.

Depending on whether you selected a single dataset (mouse cortex, human non-cortex for example) you see different views then when selecting a combined dataset (e.g. human combined). Please select an appropriate dataset from the dropdown menu on the top left.

Please contact us under [Javascript needed] mentioning this tool (ZEBRA) and the issue you encountered (and under which conditions it appeared).