Forum:Use Genetalk To Find The Second Patient With A Rare Genetic Disorder

Question

[peter.krawitz]

To solve the molecular genetics of an unknown genetic disorder usually two or more unrelated patients with the same phenotype and mutations in the same gene are required. However, as most of these disorders are so rare it is really difficult to find enough patients to screen for.

Now more and more people are using GeneTalk to find a colleague with a patient that has a rare mutation in the same gene. We think that is a great idea and helped by adding some de novo variants from the group in Nijmegen, that they suspect to be highly likely disease causing. Check out the recent annotations and comment on them! We are curious who is gonna be the first to find a second patient on www.gene-talk.de!

Answer 1

[karl.stamm]

What kind of computational throughput does the proposed GeneTalk have? Each of my patients has more than 10,000 mutations, is it feasible to search for so many variants?

Additionally, I have reservations about sharing human health information. Too many SNPs could identify a patient, especially if you are familiar with my field of research.

[peter.krawitz]

An exome will yield 20.000-30.000 sequence variants depending on the exact definition of the target region and the population background. If you are looking for the causal mutation of a rare monogenic disorders, you usually start by filtering out genotypes that are more frequent in the population than the disease itself. Additionally one usually reduced the variants to a set that has an predicted effect on the protein level and that would fit to the suspected mode of inheritance. With these filter settings one usually has to deal with a few hundred "family specific" variants. If more than one family member or additional unrelated patients are available for the analysis the number of candidate variants may be much lower, sometimes only several mutations in a few genes. For these highly likely candidate mutations it makes sense to check, whether there is anyone in GeneTalk also interested in these genes. Patient data has to be handled with care and every user is required to do so in GeneTalk. However, if a user adds an annotation in GeneTalk that he or she identified a certain sequence variant in a patient with a certain phenotype that information by itself will not help to identify the individual of with the sequence variant: Usually the phenotypic description is not specific enough and with more than 7 billion people on the planet even a specific mutation in the 50MB exome is unlikely to be unique. However, the identifiability of a person based on its genomic sequence increases if combinations of rare sequence variants are posted. However in GeneTalk this wouldn't be called a sequence variant specific annotation. It would rather be sharing the entire set of sequence variants or a subset with another user. This is only possible in GeneTalk if the patient gave his or her consent for it.

[karl.stamm]

I think what you're saying is that GeneTalk does not share sets of mutations in individuals, and concerns itself with relatively anonymous single point mutations; thereby avoiding the identifiability issue. I see also the PersonalExomeProject attached is working towards a database of rare/random/benign background noise for such data. Very nice

[DG]

I think the idea is great, assuming you are only sharing a handful of potentially interesting variants OR population level data about variant frequencies across multiple projects. That could also be useful. I know I run in to some unique restrictions on sharing genetic data from my projects due to stricter privacy laws in my jurisdiction than anywhere else in the country in terms of "health information"