Deanna Church
National Institutes of Health (NIH)
9000 Rockville Pike
Bethesda, Maryland 20892

Dear Deanna Church:
I read with interest the article about your work (The genome finishers, Elie Dolgin, NATURE vol. 4662 no. 7275 page 843).  I have a question you might be kind enough to answer.  It comes as a surprise to most people, but the fertility of a couple is determined by their kinship as well as the kinship of their parents and apparently for a few generations back.  If you want to have children, marry a cousin, not a first cousin or maybe second but as close as you can get past second cousin.  This may be a bit of an oversimplification, but the bottom line is about right.  The fact that we all marry strangers in this day and age appears to be causing a very serious fall in birth rate, so severe that it has been thirty years since the developed world has had a birth rate consistent with survival.  The evidence and references are on the enclosed DVD. 

Not included on the DVD is the effect of modern fertility measures, but even though they are somewhat effective and have to a certain extent reversed the decline, we are still below replacement.  That and other information are posted on my website 

The obvious solution is to find those third and fourth cousins.  Only not many people know who they are.  One certainly does not want to go out past sixth cousins, and very few can name all their cousins out so far.  But of course there is an obvious solution.  A data base that contained enough genomic information about enough people would be able to distinguish between couples who could hope to have children and couples for whom that hope would be in vain.

The problem is that nobody is doing the studies or assembling the data to make that possible. 

There is another problem.  According to the article, the highest degree of accuracy mentioned was one part per 10,000. 

Shared DNA falls off rapidly with higher degrees of separation.  Siblings share ½ of their DNA, but first cousins share only 1/8.  Second cousins share 1/32 and third cousins 1/128 and that is where interest just begins.  Fourth cousins share 1/512 and fifth cousins 1/2,048.  You would really like to be able to go out to sixth cousins or 1/8,192.  But to be confident within one standard deviation, you need to square that number so you need an accuracy of 1/67,109,964.  For a second standard deviation you would need even more.  That would be theoretically possible, given the enormous size of the human genome, but it might not actually be possible at all. 

If accuracy is only one part per 10,000, then it seems futile.  I suspect that carrying it out to sixth cousins will be futile for some time to come.

And of course the article points out that things are not even that simple.  It isn’t just SNP’s.  There are stretches of millions of base pairs deleted, inserted, inverted and duplicated.  Millions does not seem like much compared with billions of base pairs in the genomes, but even by the time you are looking for 5th cousins, that is more than significant.

So what I want to know is, given the current state of the art, just how far out can you detect cousins? 

I would like to post your response on my website.  I will post this letter there in case somebody else has in interest.

Thank you.


M. Linton Herbert MD and

There have been 3,136 visitors so far.

Home page.