SCENIC was run on all the datasets using the expression matrices provided by the authors (downloaded from GEO or the authors website), including only the cells that passed their quality control, and the default gene filtering for GENIE3 (which resulted in 12-15k genes). The standard SCENIC workflow was run on all datasets (the version at the time of publication is available as supplementary file, updated versions can be found at http://scenic.aertslab.org). A more detailed description of the datasets and the any peculiarities for each analysis are available in Supplementary Note 1. Here we provide a brief description of the datasets:
Mouse cortex and hippocampus (Zeisel et al.9 (link), GSE60361): single-cell RNA-seq of 3005 brain cells of juvenile mice (21-31 days old). It contains the main cell types in hippocampus and somatosensory cortex, namely neurons (pyramidal excitatory neurons, and interneurons), glia (astrocytes, oligodendrocytes, microglia), and endothelial cells. Expression matrix units: UMI counts.
Human neurons (Lake et al. 11 (link)): single-nuclei RNA-seq of 3083 neuronal cells from a normal human brain (retrieved postmortem from a 51-year old female, from six different Brodmann areas). Expression matrix units: TPM.
Human brain (Darmanis et al.36 (link), GSE67835): scRNA-seq from 466 cells from adult and fetal human brains. The fetal samples were taken from four different individuals at 16 to 18 weeks post-gestation. The adult brain samples were taken from healthy temporal lobe tissue from 8 different patients (21 - 63 years old) during temporal lobectomy surgery for refractory epilepsy and hippocampal sclerosis. Expression matrix units: logged CPM.
Mouse oligodendrocytes (Marques et al. 37 (link), GSE75330): scRNA-seq data of 5069 cells from the oligodendrocyte lineage. Cells were obtained from several different mouse strains and isolated from ten different regions of the anterior-posterior and dorsal-ventral axis of the mouse juvenile and adult CNS; including white and grey matter. Expression matrix units: UMI counts.
Oligodendroglioma (Tirosh et al. 38 (link), GSE70630): scRNA-seq expression profiles for 4347 cells from 6 untreated grade II oligodendroglioma tumors with either IDH1 or IDH2 mutation, and 1p/19q co-deletion. Only the tumoral cells were used for the analysis (selected by the authors based on CNV profile). Expression matrix units: log2(TPM+1).
Melanoma (Tirosh et al. 13 (link), GSE72056): scRNA-seq of 1252 melanoma cells from 14 different tumors. These include only the cells that are labeled as malignant by the authors, based on their CNV profiles. Expression matrix units: log2(TPM/10+1).
Mouse retina (Macosko et al. 39 (link), GSE63472): scRNA-seq data of 44808 cells obtained through Drop-seq from mouse retina (14 days post-natal). Expression matrix units: log((UMI counts per gene in a cell/Total UMI counts in cell)*10000)+1)].
Embryonic mouse brain (10X Genomics): Chronium Megacell demonstration dataset containing 1,306,127 cells from cortex, hippocampus and subventricular zone of two E18 mice (strain: C57BL/6).