I've been following bioconductor's OSCA book for some scRNAseq analysis (smart seq2 protocol, no spike-in). After QC, normalization (deconvolution) and transformation (log2, pseudocount 1) upon plotting the mean/variance of the genes, I noticed that the plot looked very much like a negative parabola:
This isn't what I'd seen in the OSCA examples. I was wondering if this is a cause for concern, or if I should continue forward into feature selection.
Thank you for your time! Let me know if I should provide any more information.