In the news
Find related articles on
Gene expression databases linked to personal genomic data
Sources: Science Insider, Nature paper
A new paper in Nature reports a mathematical method to link an individual’s DNA sample with the correct anonymised data in public research databases.
However, this has not aroused responses such as those to the a similar announcement in 2008, which prompted major bodies including the US National Institutes of Health (NIH) and the UK Wellcome Trust to withdraw public access to data (see previous news).
Gene expression data can reportedly be linked to the corresponding DNA via the new method making it theoretically possible to identify an associated individual sample and medical data. However, according to Science, the NIH has already said it is not concerned, saying "there may be no way to protect privacy" completely for individual genetic data and that ensuring people understand this is likely to be key in the future.
Comment: In some respects, concerns about the security of genetic data in large-scale research databases are apparently justified; there are, in theory, ways to access them at some level, albeit using highly sophisticated tools. But who would want to? It seems sensible to view this possibility with much less concern than, say, a situation such as an employer being able to access genetic test results for an individual employee.
Photo credit: freeimages.co.uk
Comments (1)
Evert (April 17 2012, 11:07 pm)
Sorry, even if I did not read the article yet, only the summary, and would not fully understand it if I did, I did understood the following. Data from 2 databases can be combined and thus it can be found that data under pseudonym 1 in database a relate to the same individual as data under pseudonym 2 in database b. So what? That was done with probalistic linking already using non genetic data. It does not mean that the identity of the individual is revealed. If I visit a conference and have lunch there and a glass of wine at the reception, using my unique fingerprints one will be able to know what I ate at lunch and that I had white wine at the reception. Stil not that such person was me (if the conference was well attended etc, etc.). I very much appreciate the common sense approach of the Comment, but there is this other common sense remark to be made first.
All this does not mean that I underestimate the problematic aspects of 'anonimity' of research data. But that is becasue in theory it can be possible to link phenotype data in research databases with data available in the public domain, such as on social media. Hence my maxim that the quest should not be for complete anonimity before data reach the research domain but that data should be completely safe once there (and e.g. not leaked to that employer mentioned in the Comment).
I have described that more fully elsewhere in my paper Patient data for health research,
Evert-Ben van Veen
All this does not mean that I underestimate the problematic aspects of 'anonimity' of research data. But that is becasue in theory it can be possible to link phenotype data in research databases with data available in the public domain, such as on social media. Hence my maxim that the quest should not be for complete anonimity before data reach the research domain but that data should be completely safe once there (and e.g. not leaked to that employer mentioned in the Comment).
I have described that more fully elsewhere in my paper Patient data for health research,
Evert-Ben van Veen