Visual Phasing Part 4 - Reconciling Matches

(00:00):
You’ve assigned all of your segments to grandparents and now you’re ready to start using that to find out where other matches fit in. But wait, you might be wrong.

(00:18):
Howdy, I’m Andy Lee with Family History Fanatics, where we help you understand DNA. Climb your family tree and write your family stories along the way. This is a segment of DNA. Be sure to subscribe to our channel and click on that bell if you wanna be notified about upcoming episodes. If you’ve been following along and trying out yourself in the first three videos, we’ve been able to create this. We’ve been able to assign the segments of these three siblings chromosomes to each one of their grandparents. And if you stopped right here and went forward, you might think that you are found the success that everything’s exact and ready to go, and you’d be wrong because sometimes there’s things that just don’t quite line up. In the first three videos, I talked about principles that needed to be followed as you’re going through visual phasing. And today’s video is no different, but there’s only one principle for reconciling matches, and that is that all matches must be reconciled.

(01:17):
Now what that means is, is there’s got to be a reason why things don’t line up. Now, there can be various reasons. This could be because your genealogical tree is incorrect. It might be that there was an npe along the way that you didn’t know about, that DNA is just now revealing. It may be that your visual phasing was wrong, you might have done the switch twice and that messed up where everything is lining up. There can be any number of reasons, but the important thing is, is that for every match that doesn’t line up, you need to go back and reevaluate that and determine is there a way that I can make this line up correctly? So in our example, we’ve had two matches, one on the paternal grandfather’s side and went on the maternal grandmother’s side. And we wanna take another look at those two matches.

(02:07):
Starting with the paternal grandfather’s side, there was a small segment in here that matched person B from 1 27 to one 30. Now we have said that orange was the paternal grandfather, but we see that, hey, there’s not a match up here and there’s not a match down there. So orange really can’t be the paternal grandfather in this area because it would have to match all three people, but that wouldn’t support what we already did as far as our recombination points. And I can understand that right now. But that’s something that now we need to reconcile. Let’s look at our maternal grandmother match. Now, the maternal grandmother match, it had from one 12 to 1 31, it had from one 19 to 1 31, and it had from one 12 to 1 31. We had said that teal was the maternal grandmother, but we can see right here that on person B, this match is overlapping where the maternal grandfather is. So this is curious because this happens to be in the same area as our paternal grandfather discrepancy is going to be. And so it’s likely that this area of our visual phasing is incorrect and we need to do something different. So step one is we want to delete the assignments of those non reconciled segments, and that’s going to be these ones right in here.

(03:44):
We want to go in and we wanna delete those, and we’re going to actually start over with our match data just like that. They’re gone. Now, it’s important to remember that our matches have to align with the recombination points and the segment data and vice versa. Originally what we did is we created these segments based on the recombination points. Now we’re actually going to go a little backwards. We’re gonna use those matches to create these segments on these areas that our suspect. So step two is we want to assign non reconciled segments based on matches. Right now, we’re still going to use these same four colors and they’re gonna represent our paternal grandfather, mother and maternal grandfather and mother, but we’re going to assign things based on the matches to see what we get. So let’s start with our paternal grandfather match. He matches B at 1 27 to one 30. So in this area, on B, he matches. So we can color that orange, and we know that he doesn’t match person A and person C. So we want to color those. The other one that’s gonna be the maroon one. So if I go here and I color this maroon and I color this maroon,

(04:22):
Now I’m only doing those segments that are common with the match. I’m not crossing the recombination points, I’m not extending these down and over. I’m just coloring those segments. So now let me go to the maternal grandmother. This is match number four from our example,

(05:45):
And we have person, our maternal grandmother is teal, so we wanna make sure we select teal and maternal grandmothers from one 12 to 1 31. So that actually spans both of these matches. I’m sorry, both of these segments. So I want to color that. B is from one 19 to 1 31. So that actually is just this part right here. And C is one 12 to 1 31. That spans both of ’em. So I’m going to color that one right there. So this is now what we have. We still have these segments right in the middle that we haven’t decided on yet, but we have the other ones already. So we go onto our next step, and that is we want to check for unidentified recombination points. Points. Now what do I mean by unidentified recombination points? Well, visual phasing, just like it says is visual. You’re using a graphic to determine where things happen.

(06:49):
As we can see in, even in this one, there can be recombination points that are really close together. There can be recombination points that are even closer together, and that can start to confuse what we’re seeing visually. Now, that’s because we usually use a much smaller graphic. This graphic is representing several thousand snips along this chromosome, and this graphic is only maybe a thousand pixels wide. So it’s not actually showing all of those. It’s only showing one out of every 20 of those we want to go through and we want to see what missing recombination points will be. Now this is where looking at the numbers of mega bases starts to be important. So for instance, we have a couple of things is we already had identified that one, 18.9 was the recombination point based off of what GEDmatch was reporting. When we’re looking right here, we see that hey, this one 12.8 and this one, 12.4, it shows up a couple times, and then there’s this one 19.3, which is close to one 18.3.

(08:00):
But if we’re talking about recombination points that are really close together, it might be that this is just not showing up the right way because of how close that is together. So we’re going to go to GEDmatch. We’re gonna do the one-to-one comparison, except we’re going to do the full resolution and we’re gonna zoom in on this area between about 110 and a hundred and twenty, a hundred and twenty one mega bases. So here we are, and this is going to be all of the data. So each pixel represents a snip. This is the starting point at 110. So we count over two tick marks. That’s 112 million, and that means this next one is gonna be 113 million. So one 12.31, 12.5 or six. That’s all right in this area right here. What we’re doing now is we’re taking a look at these areas and seeing is there anything that is odd where there may be a miscall or a single red or something like that that might give us a clue that this has a recombination point.

(09:07):
And right now I’m not seeing anything on here that’s telling me that there’s a recombination point. It looks like it’s pretty consistent as far as the yellow green on both AB and ac. There’s no real overlap between those two. We can see on the BC that there are a couple of red tick marks in there as well as the yellow green. And again, it is not consistent with the AC or the ab. So I’m gonna say that that one, 12.81, 12.3, there’s not an extra recombination point in there. Now if I go down to the one 18 to one 19.3 area, well this one 18 is right here, the one 18.3, and that’s where jet batch is saying, Hey, this starts. Now, if you remember, I said that no matches. Those red regions, they start and they end on a red marker. But within it you can see that there is lots of yellow and green markers interspersed throughout.

(10:10):
So the reds don’t necessarily begin and end on a red marker. They may begin and end before and after that red marker. You just don’t know for sure because it’s just the yellow and green, which looks exactly like a half match. And that’s because humans just share a lot of DNA. So if I look past that, and again, here is one 20 and this next tick mark is gonna be one 19. So I’m looking between one 19 and one 20, I can see that there might be some recombination points that I’m not seeing. Now, originally we called this recombination point as a C because these two right here, they almost aligned. And the common one there is C. But now as I’m looking at this, I can see, hey, this might be a fully identical region. That’s a really small, fully identical region. I can tell just by looking at this that this is only about 50 snips and it’s maybe a 0.6 mega basises.

(11:29):
So from a centimorgan standpoint, it’s probably 1 centimorgan or less as far as a fully identical region. So it’s not gonna show up on any of your match data, even when you’re lowering the threshold really low. As I look down, I can see that, hey, that sort of lines up near to where this actually starts going from green to yellow, and that might be where this red actually ends. Instead of the red ending right here, we might actually have the red shifted over and it ends right here. We just don’t have another no match snip, even though this whole segment might have all been part of this. So let’s take a look at that and see if that solves our problem. Which brings us to step four. Sometimes a recombination point when they’re really close together like this, we’re gonna call it as a point C, but really what it is is it’s a double recombination point, an A and a B that are right next to each other, and that’s what makes it look like a C.

(12:40):
Let’s change this to an A and a B, and then let’s C what our colors change to. So starting with person A, and there’s a recombination there, which means that this purple has to change to a orange. And so I’m going to color this first part orange up in here, and since there’s no other A that lines up right with the other orange of that segment. Next, let me go on to B now, because I don’t know what either one of these colors are. I don’t know whether that orange changes or whether that blue changes. So I’m gonna go to C and see if we can use that to figure it out. Well, C, there is no recombination point here. Now remember before there was a recombination point, but now we’ve removed that recombination point, so I’m going to make this whole thing maroon, and now we can go back up and we can see that BC is opposite of each other. So if this is maroon and teal, then this has to be orange and purple. So let me go back and let me color my orange in and then let me color my purple in.

(14:11):
Now I can go and do a quick check. So A and B, they are half matched. They have to share one color and they share orange A and C. They have to share one color and they share teal and then B and C, they can’t share either color and they don’t. So this solution works. So after I’ve done that, now I wanna go to the next step, and that is to validate this using that full resolution. I just assumed something and I tried to change something and see if everything works, and this happens to be where those matches line up. Now I want to validate this, and to do that, I’m gonna go back to this full resolution. And again, I’m focused on this area between one 18, and what I said is, is that there’s A A and a B. So can I assign an A and a B in this area?

(15:05):
Now this example can show where something’s really tricky because where half matches start and where half matches end and full matches start and full matches end is a little fuzzy. So we need to think about that as we’re doing it. Now, what we did say though is we did say that there’s not a C recombination here, so I shouldn’t be looking for common things between AC and bc. What I need to look for is common things between AB and ac, and I need to look for common things between AB and bc. So let me first assume that this half match is not starting until much later. So this is all part of the no match right at the beginning here. At this point, I have A, B and B, C. So to me, that’s looking like a nice little B. Where a start is next is I want to see where the end part is.

(16:07):
The next part has to be an A. And if we’re looking at this full match up on a B, it ends almost right about the time that BC does. But just like with a B, if we assumed that this AC really had more yellow as it goes further along, then that might indicate where the other match is. So we could say that A, B and AC have a common point right along here, and I’m gonna call that one a. Now this, as you can see, just looking at this, it could be kind of hazy and you could make another call that way. But remember, principle one on reconciling is that all matches must be reconciled. So no matter what you choose or how you choose it, it has to match with all of your matches. So if you come up with a different solution, you need to go back and resol and then check against your matches.

(17:16):
And if your matches don’t line up, that can’t be the right solution. So what I’ve done now is I have gone and I’ve put in two new recombination points, one assigned to B and one assigned to A, and I’ve changed the numbers of what they are from that one, 18.9 to one 19.3, and about one 19.8. When I do this, all of those matches are reconciled. They all line up with what my known information is and what this chart shows. Now, if you’ve tried this and you just can’t get it to work, then it’s very possible that your match from a genealogical standpoint is not related to that person, not related to that grandparent. And so you may need to go back and validate your genealogical evidence to show whether or not it is really through a different person. And that leads us to step six.

(18:12):
You may think that you’re done right now, but you’re still not done. In fact, with visual phasing, you’re really never done. As with all genealogical evidence, there can be new evidence and every new match you have is new evidence that needs to be reconciled with all of the old evidence. So as you find new matches, you need to compare them to your visual phasing and see do they line up with what I expect? If not, you gotta go through the reconciliation process again, which may mean changing your graph, which means re comparing it to other matches. And if nothing works, maybe you need to go back and check your genealogical records and see whether or not that person’s really related in that way. Now, the reason why I chose this is because I happen to have information on my paternal grandfather and my paternal grandmother, and so I can actually compare what I’ve done here with what the DNA actually shows.

(19:14):
So if I pull up my person with the grandfather of person A, you can see that it lines up nearly perfectly with that blue bar. Now, again, this is probably due a little bit to scaling issues, but that’s exactly what I would expect. That blue bar is right over where the paternal grandfather is. I look at person C, and again, that blue bar is exactly over where the paternal grandfather is, which means when I pull up my grandmother, it’s going to be the opposite and those other areas are covered, and I pull up person C who didn’t share anything with the paternal grandfather once we solved it. And that’s exactly what the DNA shows. So this is an example of how visual phasing is able to recreate the DNA. And because I used people who I have the grandparents’ DNA, I can validate that this process works.

(20:12):
And that is the end of this series on visual phasing. Remember, each chromosome is separate and all of your matches must be reconciled. If you keep those two in mind as well as the other principles and follow the steps, then you’ll be able to help recreate some of the DNA from your grandparents. Now, if you want to do this with four or five or six siblings, that’s perfectly fine. I find that it’s a lot easier to just do it with three. So if you have four siblings, and I’d do it in two sets of three where you actually have two people that overlap. The great thing about that is is their solution should be the same for both ways that you do it the same way with five and six and even more siblings. If you have any questions about visual phasing, put it in the comments below and also join the visual phasing Facebook group where there’s lots of people who have experience with visual phasing that can help answer your questions and help you solve any problems that you come up with. If you like this video, give it a thumbs up. Make sure you share it with all your friends.