Excess IBD Regions

by Kevin Alan Brook

As genetic genealogists, we know we need to verify our matches via triangulation to confirm that they are identical-by-descent (IBD) from a common ancestor. Triangulation means every member of a match cluster (triangulation group) has to match every other member who is expected to have overlaps with them (depending on what portions of it they inherited) across their particular portions of the same ancestral segment on the same chromosome, and this ancestral segment has definable start and end points but not every holder of the segment will have inherited the whole thing.

We should also try to phase the inheritance of matches between parent-child pairs. If a segment is above 3.5 centimorgans (cM) long and phases from a parent to a child, it is usually (though not always) valid. It's best to stick with phased segments above 6 cM for high confidence, and those segments must also belong to a triangulation group with a total of at least 4 segment holders who are not very closely related to one another.

Clusters of false matches fail to triangulate and appear in a person's match list due to a phenomenon called identical-by-chance (IBC). There are certain areas of DNA that are notorious for large numbers of matches that are either false or ancient. When a segment appears solely within an Excess IBD Region, it should probably be disregarded as it will be more likely than not to be a time-waster leading to the discovery of false matches.

"Relationship Estimation from Whole-Genome Sequence Data" by Hong Li, Gustavo Glusman, et al. published in PLOS Genetics on January 30, 2014 in its volume 10, number 1, identified the following positions with widespread Excess IBD Regions:

I've bolded the segment areas on chromosome 1, 15, 21, and 22 as especially problematic. I usually get overwhelmed with false matches in those areas.

The segment area on chromosome 9 provides very old matches as this is a well-known "cold spot". This is true even when the segment begins at an earlier position like 33 million. There may indeed be a common ancestor, but he or she lived very far back in time, sometimes more than the range of 20 generations that direct-to-consumer autosomal DNA tests are limited to. (By contrast, most matches we find on more normal segment areas are related to us within the past 16 generations and will be 14th cousins or closer.)

I have used GEDmatch since 2015 and I've observed several more positions that frequently give trouble.

That is not to say that all false matches will appear on the above areas. A false match can appear just about anywhere.

Some types of false matches include:

  • Genetics of ethnic groups around the world
  • Small Ethnicity Percentages are Sometimes Real - another essay by Kevin Alan Brook

    Khazaria.com receives monetary commission payments from Family Tree DNA and Viglink from sales generated through links.