DeVeale B, van der Kooy D, Babak T. Critical evaluation of imprinted gene expression by RNA-Seq: a new perspective.
PLoS Genet 2012;
8:e1002600. [PMID:
22479196 PMCID:
PMC3315459 DOI:
10.1371/journal.pgen.1002600]
[Citation(s) in RCA: 190] [Impact Index Per Article: 15.8] [Reference Citation Analysis] [Abstract] [MESH Headings] [Grants] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 07/27/2011] [Accepted: 01/18/2012] [Indexed: 01/05/2023] Open
Abstract
In contrast to existing estimates of approximately 200 murine imprinted genes, recent work based on transcriptome sequencing uncovered parent-of-origin allelic effects at more than 1,300 loci in the developing brain and two adult brain regions, including hundreds present in only males or females. Our independent replication of the embryonic brain stage, where the majority of novel imprinted genes were discovered and the majority of previously known imprinted genes confirmed, resulted in only 12.9% concordance among the novel imprinted loci. Further analysis and pyrosequencing-based validation revealed that the vast majority of the novel reported imprinted loci are false-positives explained by technical and biological variation of the experimental approach. We show that allele-specific expression (ASE) measured with RNA–Seq is not accurately modeled with statistical methods that assume random independent sampling and that systematic error must be accounted for to enable accurate identification of imprinted expression. Application of a robust approach that accounts for these effects revealed 50 candidate genes where allelic bias was predicted to be parent-of-origin–dependent. However, 11 independent validation attempts through a range of allelic expression biases confirmed only 6 of these novel cases. The results emphasize the importance of independent validation and suggest that the number of imprinted genes is much closer to the initial estimates.
Typically both copies of mammalian genes are expressed, but in some cases, “imprinting” restricts expression to the maternal or paternal copy. Having two copies of each gene is considered advantageous since in enables compensation when one does not function properly. Why imprinting evolved and its utility to each sex is widely debated, and having a complete catalog of imprinted genes and their functions is essential for fully characterizing this phenomenon. 25 years of screening has revealed about 130 imprinted genes, and the slowing rate of discovery suggests that we are reaching saturation. Two recent studies based on high-throughput sequencing of RNA reported more than 1,300 imprinted genes. To understand the basis of this paradigm shift, we first attempted to reproduce these results. Unable to do so, we performed additional analyses that show that most of these discoveries are due to noise in the experimental approach and assay. We remedy this with new methods that account for this noise and applied them to identify 50 novel putative imprinted genes. These methods will be useful for identifying genuine novel cases of imprinted expression as this type of screening approach becomes broadly utilized.
Collapse