My X-chromosome headache

It all began when the data of one distant cousin messed everything up by not fitting into the visual phasing picture I’ve previously created for my father and his siblings. It ended with a discovery of a so-called hidden recombination (more than one actually!), a situation when your visual phasing results turn out to be entirely different from what they initially seemed due to very close crossover positions of two or more siblings.

I’ve read about hidden recombinations in the Visual phasing working group on Facebook before, but I’ve never come across one myself until recently, and I would like to share that discovery with you. But first, let me give you some background information about my paternal side of the family for the full picture. 

My father’s family is of Russian ancestry including Native Siberian admixture, which is typical for long-established residents of Siberia, who settled there in the late 16th century, and occasionally intermarried with local populations. Despite having a large number of Russian and Finnish matches, all of my father’s genetic cousins are really distant, and neither their family trees nor mine go back far enough to find a MRCA.

Despite all the great tools out there, creating genetic networks for my paternal side turned out to be a rather difficult task, because nearly all of those distant cousins share less than 20cM between themselves and my father. And since no one of my father’s known relatives beyond the first cousin level has tested, I was not yet able to map any of my paternal chromosomes back to each of my great grandparents, let alone distant ancestors.

However, after comparing my father and his younger brother via the One-to-one X-DNA comparison tool at GEDmatch, I realized that thanks to the way the two of them matched, it was possible to sort their X-matches into two different groups, and then hopefully map the segments of both groups to a specific ancestor. (The original version read “segments of one group” referring to the ancestor who provided the unrecombinated copy of the X.)

According to this image my father and uncle P. share only one segment of about 28cM on the X-chromosome. This means that except for a stretch between 138.8Mb and 149.5Mb, both of my paternal grandmother’s X-chromosomes are represented in two of her sons.

One of my paternal grandmother’s two X-chromosomes is a unrecombinated copy she inherited from her father, which he in turn received from his mother, my great great grandmother Varvara Kachulina. Thus, Varvara’s DNA was represented to a large part in either my father’s or my uncle’s X-chromosome, meaning that there was a 50% chance for me to have inherited such a large portion of my great great grandmother’s DNA!

I got really excited at that thought, and immediately set out to look for my father’s and uncle’s X-matches in all databases. However, the first solid X-match I came across yielded a big surprise! Distant cousin Alexey matched my father (and me), two of his siblings and his aunt on the X-chromosome with the cM amount varying between 21cM and 48cM. The thing that struck me right away, however, were the positions of the DNA segments they shared – they just didn’t seem to make any sense!

Here is the 23andme X-chromosome comparison between distant cousin Alexey and my great aunt, uncle, aunt and father:

According to the One-to-one X-DNA comparison my father and uncle didn’t share any DNA until 138.8Mb. According to the 23andme comparison, however, Alexey shares two segments with each of them before 138.8Mb. Alexey, my father and my uncle all have only one X-chromosome, and logic tells me that person A can’t match person B and also person C, when B and C don’t match each other. And yet, I was presented with this data. Something was definitely amiss here!


I’ve tested three of my father’s siblings for Visual phasing, but I initially preferred to work with my father and just two of them, and to use the segment data of the third sibling only when one or the other recombination position remained unclear. When I was phasing their chromosomes for the first time, I only used the data of my father, aunt L. and uncle P. (the LDP Visual phasing project).

Visual phasing of the X-chromosome is usually a lot easier than that of the autosomal chromosomes, and it’s often recommended for beginners, because you are working with the maternal copy only, and therefore, only two grandparents. I remember finishing it really quickly, and back then I didn’t see any reason to doubt the outcome.

The Visual phasing of LDP looked super easy – and it still does, right? But let me jump ahead and tell you that instead of three crossovers my aunt L. turned out to have only one in the end. Surprised? I certainly was!

As you can see, this Visual phasing result doesn’t correspond with cousin Alexey’s DNA data. After staring at the screen for a while, I remembered group member Adam N.’s remark in the Visual phasing working group on Facebook about a situation when two recombinations of two siblings could look like one recombination position for the third. Could that be the case here, too?

I decided to phase the X-chromosomes again, this time leaving my father out of the picture and adding aunt G. instead (LGP Visual phasing project):

The LGP phasing result showed only two crossovers for aunt L. Her third crossover at around 33Mb turned into the second crossover for my uncle at 34Mb!

The same was true when I phased the X-chromosome again using my father’s and both of his sisters’ data (LGD visual phasing project). Aunt L.’s crossover at 33Mb turned into my father’s crossover at 32Mb.

I went back to the One-to-one X-DNA comparison tool and lowered the cM thresholds. Sure enough, there was another segment of less than 5cM that my father and uncle shared between 32Mb and 34Mb. Segments below 7cM are considered unreliable, and I usually try to avoid using them, but this one had to be real. The omission of this small segment was the reason why my father’s and uncle’s crossovers at 32Mb and 34Mb, respectively, looked like a single crossover for my aunt L. in the LDP Visual phasing project.

However, the new phasing result still didn’t align with Alexey’s data. So in the next step, I compared all four siblings together:

The ownership of all recombination positions was clear, except for the one in the middle at around 95Mb. I hesitated, because it was indicating that there were three crossovers for three siblings at nearly the same place – something that I thought was very unlikely. Also, the chromosome browser at 23andme didn’t show a break around 95Mb for aunt G. and Dad, which only added to my confusion.

On the other hand, there was Alexey’s data that didn’t fit in with the results. In addition, I’ve made another observation suggesting that a mistake was made during the Visual phasing process. According to the phasing result both of my paternal grandmother’s copies were fully represented in her four children. Since she and her sister shared an identical paternal X-chromosome (the chromosome their father Ivan Permyakov received from his mother Varvara Kachulina), my great aunt should also have been matching fully with her nieces and nephews along either M1 or M2. And that was not the case. According to the image below she matched them almost entirely along M2, except for a small region from the beginning until 12Mb.

I decided to post my problem in the Visual phasing working group on Facebook, and ask for a couple of fresh eyes to take a look at it. Shortly after, group member Adam N. confirmed that the phasing was incorrect, because “something funky was happening in the middle”. When I asked him, if he thought three close crossovers for three siblings were possible, he said it was “not the most common, but could certainly happen”, because only “a few things are completely impossible with regard to recombination placement”. Adam suggested three crossovers for G, P and D.

Now look at this – cousin Alexey matches everybody along M1 only! And even though he cannot be directly compared to my aunt L., because they have tested at different companies, thanks to Visual phasing, I’m able to predict that he and aunt L. will share three segments on the X:

7 Mb-13 Mb
34 Mb-39 Mb
71 Mb-111 Mb

Moreover, M1 can be assigned to my great grandmother Raisa Zenkova, because my great aunt fully matches her nieces and nephews along M2 now, which has to be my great great grandmother Varvara Kachulina’s X-chromosome that was passed down through her son Ivan Permyakov.

And since I inherited my father’s unrecombinated X-chromosome, I now know what my paternal copy looks like:

Did you notice how different my father’s and uncle’s chromosomes now look like compared to the first Visual phasing result? When I first saw how the two of them matched, I thought they both inherited a different, mostly unrecombinated copy of each of my paternal grandmother’s chromosomes. Instead, it turned out that both of their chromosome copies recombinated more than once, and even twice at very close locations, when both of them changed to the opposite grandparent!

Dad’s and uncle P. ‘s X-chromosomes version 1:

Dad’s and uncle P. ‘s X-chromosomes version 2:

Dad’s and uncle P. ‘s X-chromosomes version 3:


Be aware that your Visual phasing results may be incorrect due to hidden recombinations, and therefore, the outcome may be subject to change when new genetic cousins show up.

Close crossovers of two siblings can occasionally look like a single crossover for the third sibling. My father’s and uncle’s crossovers at 32Mb and 34Mb were mistaken for my aunt L.’s recombination point at 33Mb in the first Visual phasing project. The error was revealed when one of the brothers was replaced with another sibling.

Close crossovers for different siblings result in small segments, which can easily be overlooked at the default level of 7cM. Sometimes double checking your results by lowering the cM threshold can clarify things.

There are cases when a crossover may be completely hidden in the visual representation and its presence only suggested by external hints, as it was the case with my uncle P. ‘s crossover at 94Mb. Nothing in the three previous visual phasing projects (LDP, LGP, GDP) indicated that he had another crossover in the middle. The first hint was the four siblings comparison – there was visual evidence of crossovers for G. and D., but in order to make the rest of the comparison work, you had to assume another close recombination point for P. as well. Another clue was how cousin Alexey matched three of the siblings in the area between 92Mb and 95Mb.

Always investigate the reason behind conflicting results until they make sense. No matter how unlikely some evidence may appear to you, do not easily dismiss it. As it turns out, multiple siblings can have multiple close crossovers on the same chromosome!

Thanks to the different inheritance pattern of the X-chromosome and the visual phasing technique, paying close attention to how distant cousins match you and other family members, can provide you with valuable clues to help you map your chromosomes back to more distant ancestors, even if you previously thought you couldn’t. Distant cousin Alexey didn’t provide any information about himself, or reply to my message, and yet it was still possible to connect him to my great grandmother Raisa Zenkova. Besides, I can now work with the autosomal segments he shares with my family on chromosomes 2,9 and 17, too!