Wednesday, June 5, 2013

Time to get serious about DNA relatives

DNA tests are one of the most recent tools to aid researchers in the field of genealogy.  My own belief is that it will become increasingly (in fact, exponentially) more effective and important as the technology becomes cheaper and more widespread.  But even today, many casual DNA genealogists stop short of making truly effective use of their results.  To do so, you and your match must compare chromosomes and attempt to triangulate your matching segment with a third party.  Otherwise you are just guessing, and that's not good genealogy.

I was first introduced to this concept via a cute comic strip.  The crux of the problem is that people may share more than one line of genetic descent.  If Alice and Bob are related, a DNA test will show that.  But the relationship may be complicated: they can have multiple shared ancestors.  Many combinations are possible.  What's more, Alice and Bob may know of one shared ancestor, but their DNA connection might be through an ancestor that is unknown to one or both of them!

Congratulations, you found a DNA match.
You still haven't proved anything.
23andMe and Family Tree DNA both provide chromosome browsers to see where your genome connects to your matches.  However, neither provides a utility to triangulate segments, and both let you compare only a few matches at a time.  My solution was to meticulously download all the matching segment data into an Excel spreadsheet and do my own comparisons.  This works but it seems unnecessary in the 21st century.  Both companies should be doing this automatically and grouping matches together by shared segment.
  An example of shared segments in
  FTDNA's chromosome browser
The much newer AncestryDNA service is even worse in this regard.  It does not (yet) provide even a rudimentary chromosome browser.  I have taken to basically begging my Ancestry matches to download their raw data and email it to me so I can compare it using an excellent utility by David Pike.  I then use another tool, the Rutgers Map Interpolator, to determine a degree of relatedness (i.e. a centiMorgan measurement).

There exists a fantastic third-party website called GEDMatch that aims to make all this easier.  It accepts raw data from any of the aforementioned services and runs its own comparisons and triangulations.  Unfortunately, as I understand it, the site is run by volunteers on a shoestring budget who are currently swamped by all the incoming data: DNA comparisons are computationally intensive.  The site is not currently accepting new data, but is projected to start again in August.  I encourage everyone to make use of this tool when it becomes available and then to DONATE so they can continue to function.  I'm glad to see that they stepped up where the large companies didn't.

No comments:

Post a Comment