genome mosaicism. Modifying DNA extraction protocol alters human brain copy number estimation

Copy number (CN) mosaicism may be relatively common, and relevant to disease, but potential confounding influences of DNA extraction methods on precise CN determination have not been explored in human tissues. Results Human brain DNA extracted from cerebellum and frontal cortex (FC) showed possible widespread subtle CN changes on array CGH, mostly cerebellar losses in GC-rich regions, even when using matched FC DNA as reference. Droplet digital PCR (ddPCR) was consistent with mosaic losses. Varying the DNA extraction protocol radically altered results, essentially normalizing ddPCR CN, and abolishing differences between different brain regions. DNA samples extracted from the cerebellum with different protocols showed predictable, GC-content dependent, aberrations on array CGH, and different coverage depth variation on whole genome sequencing. Conclusions The clear influence of the DNA extraction protocol rules out mosaicism as a cause, and constitutes the first demonstration of this phenomenon in a human tissue, consistent with a previous report in rat tissues. CN estimation may be misleading after uneven DNA extraction, and care must be taken in interpreting GC-rich losses in particular.