in disease during development. Additionally, improved coverage of brain-related regions, such as the entire CNS [10], is warranted for future research. Third, a potential limitation of the FUMA model is that averaging gene expression disregards that low expression levels of certain genes can still be relevant to disease. This caveat illustrates the challenges of accounting for factors potentially inflating statistical results as well as capturing etiological mechanisms by examining cell type-specific gene expression [10]. Finally, to identify enriched cell types, we integrated human genomic findings with mouse scRNA brain expression data. Although considerable differences exist between mice and humans, we believe our choice is justified because mouse scRNA datasets cover transcripts that are missed in human single-nuclei sequencing [8]; cover more brain regions that are believed to be important in neuropsychiatric disorders than in human, such as the striatum for which no human scRNA expression data is available; and reveal key findings consistent with human data [8, 9]. In addition, gene expression data from rodents are often of higher quality, as fresh tissue can be more readily obtained and gene expression data cluster by cell type in different species, rather than by different cell types in the same species [8].
