AncestryDNA announces changes to its matching policy

From time to time users of the AncestryDNA service are confronted by a message advising that:

Our backend services are overtaxed at the moment and we are unable to retrieve all your matches. We apologize for the inconvenience, please try again later.

At other times users are told, “Something went wrong. Try reloading the page in your browser, or come back later.”

Refreshing the page often fails to work and users are forced to accept the invitation to ‘come back later’. Sometimes even the less specialised Ancestry.com site becomes overwhelmed.

AncestryDNA has more than 18 million samples in its database. It uses enormous computer power to store data and compare DNA. This is costly, of course, and where its methods produce false positives, inefficent.

Recently the company announced that it will no longer provide details of matches with only one segment where that segment is smaller than 8 centimorgans. The company gives as the reason for this new policy that

the shorter the length of the detected IBD segment (expressed in genetic distance), the less likely it is that the detected chromosome segment is truly inherited from a common ancestor.

AncestryDNA Matching White Paper Last updated July 15, 2020 – Discovering genetic matches across a massive, expanding genetic database

AncestryDNA notes that:

  • it has “…changed the [minimum] amount of DNA you need to share to be considered a match with another individual to 8 cM”.
  • from late August “…you’ll no longer see matches or be matched to people who share 7.9 cM or less DNA with you unless you’ve messaged them and/or included them in a note, or added them to a group (including your starred group).”
  • the change has been delayed “…until late August so you have time to review and determine if you want to save any very distant matches by sending them a message and/or including them in a note or group”.

Definitions, and what the AncestryDNA announcement means

A centimorgan (cM) is “a map unit used to express the distance between two gene loci on a chromosome. A spacing of one cM indicates a one percent chance that two genes will be separated by crossing over.” 

The more centimorgans person A shares with person B the closer they are likely to be related. Your chromosomes have a total length of about 7400 cM, so you share about 3700 cM with each of your parents. Two people who share 6 to 7.9 cM are likely to be about 6th cousins, that is, they possibly share 5th great grandparents.

If you are related to a DNA match then the DNA segment or segments you share with at least one other person are identical by descent (IBD); you and your match inherited the segment from a common ancestor without recombination, and the segment has the same ancestral origin for you and your match.

The blogger Roberta Estes has estimated that 18% of her total matches share 7 cM and 30% share 6 cM. AncestryDNA’s purge of matches smaller than 8 cM will reduce her number of DNA matches by nearly half.

It has been estimated that about half the matches in the 6 to 7 centimorgan range are false positives. These matches do not truly demonstrate inheritance from a common ancestor. An error might have occurred when the DNA testing company compiled the chromosome marker: the DNA variations or the series of single nucleotide polymorphisms (SNPs) .

The companies’ matching algorithms do not treat the paternal and maternal chromosomes separately. Consequently consecutive SNP results for a short segment of DNA may appear to be half-identical in two individuals when in actuality the DNA sequences are not identical because the SNPs match on opposing chromosomes or because of errors in the matching algorithms. False matches can be the result of pseudosegments (matching alleles zig-zagging backwards and forwards between the maternal side and the paternal side), compound segments and fuzzy boundaries.

International Society of Genetic Genealogy Wiki: Identical by descent – false positive matches https://isogg.org/wiki/Identical_by_descent#False_positive_matches

DNA is inherited so a match must also match one of our parents

If a person appears to match me but does not also match either of my parents, something has gone wrong in the matching process, for if neither of my parents shared the segment that appears to justify our match that person and I did not inherit the segment from a common ancestor.

The blogger Debbie Kennett found that 54% of her matches in the 6 to 7 cM range were not shared by either of her parents.

(https://cruwys.blogspot.com/2020/07/some-updates-to-ancestrydnas-matching.html)

If about half of all DNA matches in the AncestryDNA database are very small and about half these are false positives, then about a quarter of the matches in our list of matches are false positives and are not genealogically relevant.

My experience

On 15 August 2020 I had 25,434 DNA matches to individuals. With 125 of these people I share 20 Cm or more. There are 25,309 distant matches with whom I share between 6 and 20 cM. Unfortunately there is no way using AncestryDNA to count matches in the range of 6 to 8 cM.

As both my parents have had their DNA analysed I am able to look at my very small matches to assess whether they match either of my parents.

Recently I looked at my small matches where AncestryDNA used its ‘Thrulines’ algorithm to show we both share common ancestors on our family tree and share DNA.

Fan chart showing the genetic ancestors identified by AncestryDNA using ThruLines for small matches 6 – 7 centimorgans (August 2020; Fan chart generated by DNA Painter)

I have 9 matches with shared common ancestors and shared DNA of 6 to 7cM. Of these matches 8 also share DNA with my father. The 9th match has a strong genealogical connection on my Mainwaring line and I do not doubt we are 5th cousins. However, he does not show up among my father’s DNA matches so despite being cousins we do not share DNA. It is estimated that only one third of fifth cousins share DNA.

It may be that the match with my father was somehow modified by AncestryDNA’s proprietary Timber algorithm, which tries to eliminate false matches “because they are of the same ethnicity or population — meaning that they (and many others from that same population) share DNA that they inherited from a distant ancestor who lived much longer ago”.

The Timber algorithm removes matches where the segments show “identical DNA with thousands of other people at that particular place”.

Recording notes about matches in the AncestryDNA database

AncestryDNA allows users to make personal notes in their database for each of their matches. Users can record when they messaged a match, how they believe they might be related, and the details of their shared match.

part of the list of AncestryDNA matches. Those in the list where AncestryDNA’s Thrulines algorithm have identified a common ancestor are shown with a green leaf in the 3rd column – see orange arrows. Part of the notes field appears in the 4th column see green arrows.
One of my DNA matches showing my notes about the match

I looked at the 38 matches in the range of 6 to 7 cM shared DNA where I had made notes.

Of these 38 matches only one was shared with my mother, who shares 9 cM with that match.  I don’t know how we connect.

Of 19 matches for which I had made notes and messaged, the matches also recorded shared DNA with my father. Three show common ancestors. Sixteen do not show common ancestors at present, although for several I know definitely how we are related.

Eighteen matches that I had annotated do not show as sharing DNA with either of my parents. It is thus likely that these matches and I do not share a common ancestor who can be genealogically traced.

Two of these matches had previously shown up as common ancestors, a green leaf in the list and the connection between our trees identified but no longer show up with the connection between our trees. Moreover they do not share DNA with my father.

For 2 of the 18 matches my father does not show as sharing DNA but does share DNA with another relative administered by the same manager. To me this indicates a possibility that the match is genuine but AncestryDNA’s Timber algorithm has modified my father’s match so it does not show up.

Consequences

I am not concerned if AncestryDNA removes smaller matches from my match list. Based on my very small review I accept that around half of the matches do not share DNA with my parents and thus are not useful matches for genealogy.

Of the remaining matches I already try to make connections with cousins who are researching the same ancestors using Ancestry.com’s member connect facility, which helps users to find fellow users researching the same ancestor. So while the ‘Thrulines’ method of highlighting shared DNA and common ancestors on a tree is useful, it is not the only means to make connections with cousins.

The Member Connect option is highlighted by the orange arrow
When you click through to Member Connect you can see other people who are researching the same ancestor, the facts they have used and the Ancestry.com sources they have saved to their tree. Sometimes Ancestry.com does not correctly identify the common ancestor and you can ignore their suggestion.

Other matching services

Databases such as MyHeritage, FTDNA, and GEDMatch have tools that AncestryDNA does not provide. The tools are based on a chromosome browser, which helps to reveal details about the matched segment, enabling users to compare the shared segment with other matches. Without this it is impossible to be confident that the inherited segment is indeed from a distant ancestor. (See my post on Triangulating Matilda’s DNA.)

These other matching services have databases considerably smaller than that of AncestryDNA. Notwithstanding the better tools they have and the ability to see details of shared segments, I think it is best to test at AncestryDNA in the first instance because of the larger pool of matches.

Estimate of the size of the major autosomal databases in January 2020 caculated by Leah Larkin and published at https://thednageek.com/autosomal-dna-database-growth/ In January 2020 AncestryDNA announced it had past the 16 million mark; in July 2020 AncestryDNA announced it had passed 18 million tests. When we first tested in July 2016 the AncestryDNA database was less than one sixth the size.

Conclusion

The findings that a significant proportion of DNA matches are not genealogically significant seems plausible . For those small matches that are in fact genealogically sound I am prepared to miss out on these and concentrate my research efforts on the many thousands of larger DNA matches where I can be confident that there is definitely shared DNA and a genealogical connection if I could work it out – there are plenty of puzzles yet to be solved.

I will be pleased if AncestryDNA’s decision to remove small matches relieves pressure on its computing resources and improves response time for its users.

How will these proposed changes affect subscribers? I hope that AncestryDNA will provide more information about shared matches and new tools for exploring DNA data. And I wouldn’t mind at all if I never saw the “backend services are overtaxed” message ever again.

Sources

Related posts