On the hunt for more hidden recombinations

In my last post, I’ve written about discovering hidden recombinations on the X-chromosomes of my father and his siblings after comparing them to a distant cousin match. This has raised several questions: Are there more? Where are they? How many of my previous Visual phasing results are incorrect because of hidden recombinations?

Visual phasing is still a fairly new technique – it was developed independently by Kathy Johnston and Randy Whited some time in 2015. However, the greater public became aware of this methodology only more than a year later, thanks to Blaine Bettinger’s blog post in November 2016. Since then its popularity grew among genetic genealogists around the world, many of whom have shared their experiences in the Visual phasing working group on Facebook.

While the Visual phasing methodology is the same for all, the results are unique for everyone. And so are the situations the group members find themselves in. Some have only two siblings to work with, but still succeed, thanks to other close relatives. Others lack close cousins, but manage to make the best out of their stranger matches at GEDmatch. And then there are also several lucky ones, who have the opportunity to work with four or more siblings, and reconstruct nearly complete DNA profiles of their deceased ancestors. The knowledge and insight gained in this truly pioneering work is then shared with each other in the group.

One such valuable insight was the discovery of hidden recombinations, a situation when two close crossovers of two siblings looked like a single crossover for the third, or when close crossovers were not visible in the GEDmatch One-to-one comparison images at all, because they “canceled” each other, and their presence was only suggested by cousin matches or additional siblings. Hidden recombinations were mentioned many times in the past two years, but I’ve come across them only a few months ago for the first time. Ever since, I’ve been on the lookout for more, and have encountered two other cases of hidden crossovers in my maternal Visual phasing project.


Chromosome 21 is the shortest chromosome, and a lot easier to phase than the others. My original Visual phasing result looked like this:

However, this result was challenged by distant cousin Michael on 23andme. The Relatives in common tool revealed that Michael was a match on my maternal grandmother’s side, and I’ve traced nearly all of her lines back to the first settlers on the Volga in 1767. His last name pointed to a Scottish or Irish origin, but he quickly confirmed his mother’s Volga German roots, and shared her maiden name with me. This provided a first clue to our genetic connection, and after reading the entry about her on the Find A Grave website, I was able to connect her parents to the 1873 family list of Katharinenstadt in Russia. I then followed their family lines back through several censuses until I’ve found our most recent common ancestors, my gggg-grandparents Leonhard Meis (b. 1794) and Anna Maria Nürnberger (b.1797). That makes Michael my maternal grandmother’s 3C, related to our family via Heinrich Antoni, my mother’s and her siblings’ maternal grandfather.

Michael has tested at 23andme only, and didn’t upload his data to GEDmatch, FTDNA or MyHeritage, so that I couldn’t compare him directly to my aunt. However, my aunt’s granddaughter E.C. happened to share one segment with Michael on chromosome 21.

Michael should be matching my family only on those areas of the chromosome that were previously assigned to Heinrich Antoni. However, according to this image he matches Ottilia Arnhold as well, and that can’t be right. Endogamy can be excluded here, because the region, where Michael matches my mother, is unbroken. If they were related through two different family lines, there would be different segments for each line. Besides, I extended Michael’s family tree, and didn’t find another branch we could be related on.

Instead, the way Michael matches my family suggests that it’s my uncle, who has a crossover at 35Mb, and not my mother. Remember the situation when close crossovers of two siblings can look like a single crossover for the third? Instead of my mother, both of her siblings must have close recombinations points at around 35Mb. With this in mind, I visually phased chromosome 21 again, and voilà – Michael matches only the areas assigned to Heinrich Antoni now:

In addition, two known relatives support the theory of my aunt’s crossover at 34Mb and my uncle’s crossover at 35Mb.

Their match with T.A., a maternal first cousin:

As well as their match with another maternal first cousin I.M.:

How good it feels when things finally fall into place 😉 (I have to admit that I should have noticed these hidden crossovers earlier, if I had paid better attention, and compared the end result to all known relatives, when I was phasing this chromosome for the first time.)