Tuesday, April 7, 2015

Ancestry DNA to Share (Mis)Information With Medical Research Companies

Few things ever truly surprise me. This shouldn't have, but it did:

Huffington Post reports that Ancestry.com is quietly moving into the lucrative medical data-mining business.

This was always 23andMe's business model, but I didn't realize that Ancestry.com would try to move into this market so soon, especially given how they treat their genetic genealogy customers.

Let me explain more fully.

23andMe is a relative of Google (the family history of the companies is its own interesting drama), and always appeared to be more of a DNA data-mining venture with genetic genealogy tacked on as a teaser. Still, they always gave lots of tools to users to access their data, compare chromosomes to those willing to share them, etc.

Ancestry.com is primarily a family tree research company which has a DNA arm. They do not give users access to tools like chromosome browsers, viewing such tools as niche gizmos that will confuse their average user and which will cause more confusion (i.e., require more customer support) than tangible benefit to them.

Recently, they have begun down a controversial path of focusing on what they call DNA Circles, which their computers create by comparing family trees of people with autosomal DNA matches and hypothesizing that they are possibly related.

They also appear to be combining information from all available trees to create a sort of homogenized blended bio for each person, and use some sort of lateral networking algorithm for assigning potential ancestors.

Given that misinformation seems to propagate on Ancestry.com faster than correct information, this is a recipe for disaster (Roberta Estes describes some problems with these enhancements on her blog).

This brings me back to the article on Ancestry.com as a medical research company. The money quote for me is that "since it has been collecting ancestral data about its users for decades, it knows health information not just about its users, but about their great-grandparents and great-great-grandparents."

This is simply untrue. Ancestry.com knows about people it suspects are the users' families, based upon sometimes highly questionable user-supplied genealogical data.  The temptation to click those shaky leaves and just add them to a tree has led many, many a hobbyist astray.

In the past, the worst thing about the unreliability of online trees has been that casual hobbyists or beginners might be led astray.  Now that we are discussing these trees as though they were reliable population information for medical research, the downside of those misshapen trees could have more dire consequences. 

I certainly hope that the medical researchers are less susceptible to marketing hype than the public has been.