Behind The New AncestryDNA Feature: Amount of Shared DNA

AncestryDNA recently released a feature to your DNA match results that we call Amount of Shared DNA. It allows you to see some scientific details behind your relationship with each of your matches. You now have access to this for each of your matches.  See an example below.

New DNA Matching Details

In this example, the user shares 71 centimorgans (cM) across 5 DNA segments with this particular DNA match. The box containing these details appears when you click on the info icon (the “i”) located to the right of the confidence level. From here, you can also click the What does this mean? link to learn more about centimorgans and how they are used to calculate relationships and confidence scores for each match.

How Shared Centimorgans Are Calculated

The total number of shared centimorgans represents the amount of DNA (across 22 pairs of autosomal chromosomes) that you and your DNA match likely have in common because you both inherited it from a recent common ancestor.

As relationships get more distant, it is important to distinguish between DNA that is identical because it was inherited from a recent common ancestor (6-10 generations ago) and DNA that is identical for other reasons (i.e., due to common ethnicity or older shared history); read more on this here. The number of centimorgans we report is our estimate of the amount of shared DNA that we can reliably attribute to a recent shared ancestor rather than more distant shared history. While it is possible that you share more DNA with your match, it may be shared in very short segments that are difficult to attribute to a recent shared ancestor.

In order to identify DNA shared because of a recent common ancestor, we use cutting-edge technology called Timber, which weights matching segments across the genome accordingly. Because close matches (parent/child through 2nd cousins) share so much DNA, it isn’t necessary to weight their shared segments – so, we only use Timber for matches estimated to be 3rd cousins or greater. As a result, while reported centimorgans for close matches reflects our unweighted estimate of the amount of shared DNA, for more distant DNA matches, the reported centimorgans reflects the weighted estimate using Timber. Dr. Julie Granka provides more details about Timber in this blog post.

How the Number of Shared Segments Are Calculated

This new feature also reports the number of DNA segments you share with any particular match.  Although it provides some useful information obtained from our matching algorithm, there are a few practical details that affect this number.

First, for accuracy and speed, our algorithm chunks the genome up into 44 sections across your 22 pairs of autosomes. This chunking does not always correspond with the chromosomes themselves – for example, one chromosome may be left intact, while another may be broken up into three or more pieces. In the example above where 71 centimorgans (cMs) were shared over 5 segments, you might have a 12-cM segment and a 14-cM segment shared on section 1 of chromosome 2, (see an example to the right), a 15-cM and a 9-cM segment on sections 2 and 3 of chromosome 8, and a 21-cM segment on section 1 of chromosome 18 for 71 total cMs on 5 different segments.

Second, a single segment of DNA inherited by two people from a recent common ancestor could actually show up in our algorithm as multiple shared segments within that originally inherited segment. This is due to some inherent difficulties in chromosome phasing, genotyping, and IBD estimation methods. IBD means “identical by descent” and refers to segments of DNA inherited from a recent ancestor, as we’ve discussed earlier. Even the best available phasing pipelines have some issues with obtaining perfect phase. Because of this, an IBD segment may switch between inferred chromosome copies, and be split across multiple identified segments. GERMLINE (our IBD identification algorithm) has a process for mitigating this effect in which it allows matches to jump back and forth between chromosome copies – but it is not perfect at stitching these back together. As a result, DNA matching algorithms will frequently identify two distinct matches near one another that in fact represent one long segment of IBD. As a final note, we do not use the number of shared segments in our relationship predictions -we only use the total amount of shared centimorgans, as it has proven to provide accurate relationship estimates. You can find more details about this in our DNA Matching white paper.

So, while you might expect to have no more than 22 shared segments with your mother, for example, you will actually see many more—both because of the way that the chromosomes are sectioned in our algorithm and because smaller identified shared segments are not always stitched together after they are identified. Although the number of shared segments can be a helpful guide in your research, because of the caveats we mention above, we recommend focusing your attention on the number of shared centimorgans.

We hope that you will find this information useful as you explore your relationship with other AncestryDNA members.