My father's G25 results

[Ion Basescul]

I think that you shouldn't be surprised that his results are completely different from the predicted kit. I had done those too for my father before testing him and they diverge quite a bit.
That's because while we inherit 50% from each parent, what you get in those 50% can be anything out of their 100%. You don't simply get everything from one parent divided by 2, if that makes sense.

[Kaspias]

Take into account that if Kaspias used a similar Gedmatch-based method as the one I described, those are not actually your father results,(as much as you like them because they tell you something you like), they may get quite close, or they may not.

These coordinates show what exactly Chris inherited from his father but there is a factor that there are also not inherited ones. This is kind of a low SNP kit of his father. Cannot even be compared to the Gedmatch method in terms of reliability as the spreading of components are totally random in there. But not super accurate either.

[Mr.G]

I think that you shouldn't be surprised that his results are completely different from the predicted kit. I had done those too for my father before testing him and they diverge quite a bit.
That's because while we inherit 50% from each parent, what you get in those 50% can be anything out of their 100%. You don't simply get everything from one parent divided by 2, if that makes sense.

That's right.

But I have to say Kaspias is using a method (superior to phasing imo) that seems tp be pretty accurate as far as showing what you inherited from that parent.

[gixajo]

These coordinates show what exactly Chris inherited from his father but there is a factor that there are also not inherited ones. This is kind of a low SNP kit of his father. Cannot even be compared to the Gedmatch method in terms of reliability as the spreading of components are totally random in there. But not super accurate either.

How did you do those coordinates?

If you did not use the gedmatch help/crutch, or you know to which component each coordinate corresponds and you changed them manually, or you have a multivariate statistical tool to handle raw data.

[gixajo]

Or based on his family origins, you made coordinates mean/averages with coordinates availables in individual or pop averages datasheet, but this is even less accurate than basing you in gedmatch...

[Kaspias]

How did you do those coordinates?

If you did not use the gedmatch help/crutch, or you know to which component each coordinate corresponds and you changed them manually, or you have a multivariate statistical tool to handle raw data.

PCA is not working in same function as Gedmatch components(Getting 1% more NA in Eurogenes K13 can significantly shift your position in a PCA, but writing slightly different coordinates won't change your position in PCA). I'm creating a middle point by using a reference either according to one parent's own results or using a reference point. In Chris's case, I had to use a reference for his mother which I should create myself. I plotted all Hungarian samples and diverged them according to where they plot.



Cluster 1 is came out that Germanic shifted ones which Chris absolutely lack of so I eliminated it directly. Then took the average of 2 and 3, subtracted Chris's father coordinates by using both of it. Cluster 2 was a middle point and samples were probably from Budapest, while 3 was significantly Slavic shifted as what Alföld individuals get.

I checked the output both 2 and 3 gave me and couldn't find a meaningful difference, and prefered to use 3's output as it is more realistic. Then made a quality control process by plotting Chris in PCA with other G25 samples. Moreover, I also subtracted coordinates with each sample present in cluster 2 and 3 then plotted the output to see the margin of error. All of the outputs created their own cluster in where they plotted at the beginning. This had to mean if Chris's mother have no any exotic admixture, father's results should be in this range. The important thing in this process is finding correct references and I was able to find in the case of Hungarians because I have experience on them. On the other hand, if one would supply one of his parent's coordinates as Mr.G did, there is literally no margin of error in the process and it will show what you inherited exactly.

In this case, considering the margin of error as it is not supplied by an actual parent data, there might be some changes in the single population rank for example Romanian above Bulgarian or whatsoever. But for example, the result which Chris posted(Dacian-Slavic-Turkic) won't change because this is what a person can get who plots in somewhere there. It won't change that his father will be southern shifted etc... But the thing is it is impossible to reach such accuracy in Gedmatch even if you use your parent's raw data for phasing, which I did without even using parent data.

[gixajo]

PCA is not working in same function as Gedmatch components(Getting 1% more NA in Eurogenes K13 can significantly shift your position in a PCA, but writing slightly different coordinates won't change your position in PCA). I'm creating a middle point by using a reference either according to one parent's own results or using a reference point. In Chris's case, I had to use a reference for his mother which I should create myself. I plotted all Hungarian samples and diverged them according to where they plot.



Cluster 1 is came out that Germanic shifted ones which Christ absolutely lack of so I eliminated it directly. Then took the average of 2 and 3, subtracted Chris's father coordinates by using both of it. Cluster 2 was a middle point and samples were probably from Budapest, while 3 was significantly Slavic shifted as what Alföld individuals get.

I checked the output both 2 and 3 gave me and couldn't find a meaningful difference, and prefered to use 3's output as it is more realistic. Then made a quality control process by plotting Chris in PCA with other G25 samples. Moreover, I also subtracted coordinates with each sample present in cluster 2 and 3 then plotted the output to see the margin of error. All of the outputs created their own cluster in where they plotted at the beginning. This had to mean if Chris's mother have no any exotic admixture, father's results should be in this range. The important thing in this process is finding correct references and I was able to find in the case of Hungarians because I have experience on them. On the other hand, if one would supply one of his parent's coordinates as Mr.G did, there is literally no margin of error in the process and it will show what you inherited exactly.

In this case, considering the margin of error as it is not supplied by an actual parent data, there might be some changes in the single population rank for example Romanian above Bulgarian or whatsoever. But for example, the result which Chris posted(Dacian-Slavic-Turkic) won't change because this is what a person can get who plots in somewhere there. It won't change that his father will be southern shifted etc... But the thing is it is impossible to reach such accuracy in Gedmatch even if you use your parent's raw data for phasing, which I did without even using parent data.

I don´t think the margin of error is very distant from this one of the gedmatch based method, but it´s a very good idea.

It´s a G25 Hungarian samples PCA?

If you want to orient yourself on how a son and his parents plot in a pca of the G25 I can pass mine and my parents', basically a son plots at the midpoint of a segment whose ends are his parents. What I don't know very well, is how you calculate from a child, the distance at which their parents are in that segment.

[Mr.G]

On the other hand, if one would supply one of his parent's coordinates as Mr.G did, there is literally no margin of error in the process and it will show what you inherited exactly.

Actually, I have not gotten my mother's coordinates from Davidski yet, you used the method you described above to find my father's coordinates, as I remember (it was months ago). Now that Davidski is back in business, I will send him my mom's data too.

We can re-do my father's coordinates after I have my mother's actual one's available, if you like and if you have time.

I expect the results will be basically the same, the coordinates you already created make good, logical sense.

[gixajo]

Actually, I have not gotten my mother's coordinates from Davidski yet, you used the method you described above to find my father's coordinates, as I remember (it was months ago). Now that Davidski is back in business, I will send him my mom's data too.

We can re-do my father's coordinates after I have my mother's actual one's available, if you like and if you have time.

I expect the results will be basically the same, the coordinates you already created make good, logical sense.

I you have your coordinates and your mother´s coordinates, using your PCA method and adding some maths you could be quite accurate to calculate your father´s ones.

You can base your way in my parents and my results:


(Well not so middlepoint of a line between my father and my mother as I remembered it, sorry, but it can help)



And see results and distances.

Target: gixajo_scaled
Distance: 1.6473% / 0.01647267
50.0 gixajo_dad_scaled
50.0 gixajo_mom_scaled

Distance to: gixajo_scaled
0.02410583 gixajo_mom_scaled
0.02411292 gixajo_dad_scaled

And this is not exact but...

(Mother+Father)/2≈Son

So...

Father≈2XSon-Mother

Good luck.

[17571imre]

the method from MrG works also pretty good with this calculator:

i put my moms coordinates in the calculator and then i got this results:


Target: mh_scaled
Distance: 1.6648% / 0.01664782 | ADC: 0.25x
48.2 cb_scaled (mother)
23.8 Germanic
19.0 East_Balkan
9.0 Slavic

it gives my side that i got form my father my 23.8 Germanic saying that i didnt inherit any Germanic from my mother. and a lot more higher Balkan wich could explain why i score more of it then my mother.
i dont know how accuarte this method is but the results look pretty logical