 |
News Search |
|
|
|
Research Brief: Genes and Longevity Study Retracted
|
|
|
|
|
|
22 July 2011. Researchers are retracting a paper that was published in Science last year and covered by Alzforum. Thomas Perls and colleagues at Boston University School of Medicine had reported genetic variants that predicted exceptional longevity (see ARF related news story). Part of their analysis relied on a genomewide association study (GWAS) based on the Illumina 610-Quad Array genotyping platform. As explained in a statement released by the journal, the researchers subsequently learned of unpublished data from other research groups questioning the accuracy of the Array over a subset of single nucleotide polymorphisms. First author Paola Sebastiani and colleagues corrected the data and resubmitted to Science. After peer review, editors concluded that the new data did not meet the journal’s standards for GWAS studies. The authors then agreed to the retraction, which appears in today's issue of the journal.
The Science statement makes it clear that there was no misconduct on behalf of the authors. Does this mean other studies using the Illumina 610-Quad Array brought into question? Geneticists contacted by Alzforum said that was unlikely. There are issues with all arrays that can be controlled by rigorous quality control and reliable replication, they said. One of the standards cited by Science as being necessary for publication of GWAS is inclusion of a reliable replication sample that supports the initial observations. The authors claim that the main scientific findings are supported by the new analysis, but that the specific details are substantially changed, and they plan to publish elsewhere.
The retraction made headlines in the national media today with a piece by Joe Palca on National Public Radio, citing initial poor peer review as a reason for the retraction. In a comment to Alzforum on the original paper, John Hardy, a geneticist at University College London, wrote that the study was underpowered and predicted it would be unreliable apart from variations with large effect sizes.—Tom Fagan.
Retraction:
Sebastiani P, Solovieff N, Puca A, Hartley SW, Melista E, Andersen S, Dworkis DA, Wilk JB, Myers RH, Steinberg MH, Montano M, Baldwin CT, Perls TT. Genetic Signatures of Exceptional Longevity in Humans.
Science 22 July, 2011; 333:404. Abstract
|
|
| |
|
Comments on News and Primary Papers |
|
|
| |
Primary Papers: Genetic signatures of exceptional longevity in humans.
Comment by: John Hardy, ARF Advisor
|
|
|
Submitted 5 July 2010
| Permalink
|
Posted 5 July 2010
|
|
|
My view of this study is that it is generally underpowered and that, apart from alleles with large effect sizes, it will be not reliable. The Manhattan plot shows that only six SNPs exceeded the generally accepted level of significance of 10 -7 to 10 -8 (Figure S5), and none of these had the comforting “climbing tower” of less well-defined SNPs to indicate that they are part of a haplotype block showing association (as one of many of the examples of this, see the Simón-Sánchez et al., 2009 SNCA “climbing towers” in their Nature Genetics study of Parkinson’s disease). This does not mean the associations are wrong, but it does mean they may not be reliable. The replication study is an order of magnitude too small.
A general rule is, the more complex the statistical analysis, the less reliable the outcome, because one does not know how many other complex statistical algorithms were tried. The way to do GWAS is to do simple chi squareds across the SNPs, apply a conservative Bonnferroni (10-7) and believe...
Read more
My view of this study is that it is generally underpowered and that, apart from alleles with large effect sizes, it will be not reliable. The Manhattan plot shows that only six SNPs exceeded the generally accepted level of significance of 10 -7 to 10 -8 (Figure S5), and none of these had the comforting “climbing tower” of less well-defined SNPs to indicate that they are part of a haplotype block showing association (as one of many of the examples of this, see the Simón-Sánchez et al., 2009 SNCA “climbing towers” in their Nature Genetics study of Parkinson’s disease). This does not mean the associations are wrong, but it does mean they may not be reliable. The replication study is an order of magnitude too small.
A general rule is, the more complex the statistical analysis, the less reliable the outcome, because one does not know how many other complex statistical algorithms were tried. The way to do GWAS is to do simple chi squareds across the SNPs, apply a conservative Bonnferroni (10-7) and believe that, given no genotyping confounders. “Real” hits usually (but by no means always) have “climbing towers of less associated SNPs on each side based on the local haplotype structure.
In this GWAS, some hits are likely to be real (ApoE, e.g., which has such an enormous effect on AD risk that it is likely, and has previously been shown, to have an effect on longevity). In general, AD genes like, for example, CLU, will have such a small effect on AD risk (odds ratio 1.2) that their effect on longevity will be really small unless they have other effects on other diseases, too.
 View all comments by John Hardy |
|
|
|
| |
Submit a Comment on this News Article |
|
|
|
|
|
Home
