PDA

View Full Version : Mixed Mode Population Sharing



Ais
04-14-2018, 12:09 PM
I'm very curious about Mixed Mode Population Sharing on the gedmatch Oracles, I'm not entirely sure how they work. I imagine that they choose the two different populations sources that your DNA most closely matches in combination on that particular calculator, but are you generally considered to have actual ancestry from the secondary population areas? And is it the same for Oracle 4's 4 populations approximation?

SonsOfHerakles
04-15-2018, 05:18 PM
I'm not the best person to answer this but since nobody else has helped I'll try. When you look at the Mixed Mode you should not ignore the second population area. For example, if you look at the results below the best fitting model for my DNA (according to this specific calculator) is 61.4% Slovenian +38.6% Ashkenazi. I know that this does not match my ancestry but it is interesting to see how the calculator interprets my DNA. To me the fifth result 57.1% S_Italian_Sicilian +42.9% FIN is actually closer to my know ancestry even though it doesn't actually fit the model as well. Since my ancestry is more mixed I prefer the Oracle 4 because it helps to isolate different elements. If you are only English and Scottish I would stick to two populations.

Mixed Mode Population Sharing:

# Primary Population (source) Secondary Population (source) Distance
1 61.4% Slovenian (Xing) + 38.6% Ashkenazi (Dodecad) @ 1.67
2 59.3% Slovenian (Xing) + 40.7% Ashkenazy_Jews (Behar) @ 1.88
3 67.6% Slovenian (Xing) + 32.4% Morocco_Jews (Behar) @ 2.47
4 75.7% Tuscan (Xing) + 24.3% Finnish (Dodecad) @ 2.55
5 57.1% S_Italian_Sicilian (Dodecad) + 42.9% FIN (1000Genomes) @ 2.57

Ais
04-16-2018, 03:47 AM
I'm not the best person to answer this but since nobody else has helped I'll try. When you look at the Mixed Mode you should not ignore the second population area. For example, if you look at the results below the best fitting model for my DNA (according to this specific calculator) is 61.4% Slovenian +38.6% Ashkenazi. I know that this does not match my ancestry but it is interesting to see how the calculator interprets my DNA. To me the fifth result 57.1% S_Italian_Sicilian +42.9% FIN is actually closer to my know ancestry even though it doesn't actually fit the model as well. Since my ancestry is more mixed I prefer the Oracle 4 because it helps to isolate different elements. If you are only English and Scottish I would stick to two populations.

Mixed Mode Population Sharing:

# Primary Population (source) Secondary Population (source) Distance
1 61.4% Slovenian (Xing) + 38.6% Ashkenazi (Dodecad) @ 1.67
2 59.3% Slovenian (Xing) + 40.7% Ashkenazy_Jews (Behar) @ 1.88
3 67.6% Slovenian (Xing) + 32.4% Morocco_Jews (Behar) @ 2.47
4 75.7% Tuscan (Xing) + 24.3% Finnish (Dodecad) @ 2.55
5 57.1% S_Italian_Sicilian (Dodecad) + 42.9% FIN (1000Genomes) @ 2.57

Thanks for answering, I'm trying to understand what mixed mode population sharing actually indicates! So the calculator is giving you Ashkenazi Jewish ancestry you don't have? I'm wondering, if the secondary population doesn't give a rough region you have ancestry from, what does it tell you?

These are my top 5 results on K13:

1 94.3% Orcadian + 5.7% Kalash @ 3.04
2 94.9% Orcadian + 5.1% Balochi @ 3.11
3 87.5% Irish + 12.5% Moldavian @ 3.11
4 95% Orcadian + 5% Brahui @ 3.14
5 94.7% Irish + 5.3% Burusho @ 3.14

K15:

1 88% Southwest_English + 12% Kargopol_Russian @ 3.18
2 62.8% Orcadian + 37.2% East_German @ 3.22
3 89.5% Southwest_English + 10.5% Erzya @ 3.22
4 89.9% Southeast_English + 10.1% Erzya @ 3.24
5 88.6% Southeast_English + 11.4% Kargopol_Russian @ 3.27

And EUtest:

1 90.1% IE + 9.9% UA @ 2.55
2 90% IE + 10% West_Russian @ 2.56
3 90.6% IE + 9.4% Ukrainian-Russian @ 2.58
4 95.6% IE + 4.4% Kalash @ 2.65
5 89.7% IE + 10.3% PL @ 2.66

SonsOfHerakles
04-16-2018, 01:12 PM
Thanks for answering, I'm trying to understand what mixed mode population sharing actually indicates! So the calculator is giving you Ashkenazi Jewish ancestry you don't have? I'm wondering, if the secondary population doesn't give a rough region you have ancestry from, what does it tell you?

These are my top 5 results on K13:

1 94.3% Orcadian + 5.7% Kalash @ 3.04
2 94.9% Orcadian + 5.1% Balochi @ 3.11
3 87.5% Irish + 12.5% Moldavian @ 3.11
4 95% Orcadian + 5% Brahui @ 3.14
5 94.7% Irish + 5.3% Burusho @ 3.14

K15:

1 88% Southwest_English + 12% Kargopol_Russian @ 3.18
2 62.8% Orcadian + 37.2% East_German @ 3.22
3 89.5% Southwest_English + 10.5% Erzya @ 3.22
4 89.9% Southeast_English + 10.1% Erzya @ 3.24
5 88.6% Southeast_English + 11.4% Kargopol_Russian @ 3.27

And EUtest:

1 90.1% IE + 9.9% UA @ 2.55
2 90% IE + 10% West_Russian @ 2.56
3 90.6% IE + 9.4% Ukrainian-Russian @ 2.58
4 95.6% IE + 4.4% Kalash @ 2.65
5 89.7% IE + 10.3% PL @ 2.66

It is indicating that the best fitting model is that you are 94.3% Orcadian and 5.7% Kalash. The @ 3.04 indicates how accurate the model fits your DNA, the lower the better. That being said, we know the genetic inheritance is random and also requires interpretation. IF you have no known Kalash ancestry then you are probably not Kalash, even though its possible. The small Kalash component may represent some other similar component still present in your DNA. You have to use what you know about your own ancestry plus genetics in general to find the most likely scenario. For example, I could be half Greek and half Finnish but my best fitting results could be Albania and Croatian. Why? The calculator could read the the separate Mediterranean and Baltic components as part of of one or two more closely related groups. If I know that I am Greek and Finnish though I won't just ignore what I know in favor of a better fitting population model. These are used more as hints than answers that are to be taken with certainty. Try messing around with the oracle 4 and also looking at how you compared to different populations in the spreadsheet. Oracle, Oracle 4, and the spreadsheet are all at the bottom of the results page for each calculator. I hope that helps!

Ais
04-17-2018, 12:35 PM
It is indicating that the best fitting model is that you are 94.3% Orcadian and 5.7% Kalash. The @ 3.04 indicates how accurate the model fits your DNA, the lower the better. That being said, we know the genetic inheritance is random and also requires interpretation. IF you have no known Kalash ancestry then you are probably not Kalash, even though its possible. The small Kalash component may represent some other similar component still present in your DNA. You have to use what you know about your own ancestry plus genetics in general to find the most likely scenario. For example, I could be half Greek and half Finnish but my best fitting results could be Albania and Croatian. Why? The calculator could read the the separate Mediterranean and Baltic components as part of of one or two more closely related groups. If I know that I am Greek and Finnish though I won't just ignore what I know in favor of a better fitting population model. These are used more as hints than answers that are to be taken with certainty. Try messing around with the oracle 4 and also looking at how you compared to different populations in the spreadsheet. Oracle, Oracle 4, and the spreadsheet are all at the bottom of the results page for each calculator. I hope that helps!

I assumed the Kalash was representative of my known Indian ancestry (pretty far back!) I've just been wondering about some of my strange results, and why someone who is almost entirely British would be getting mixed results such as 83.60% English (Kent) + 16.40% Ukrainian (Ukraine) @ 1.94 and 88.10% English (Kent) + 11.90% Russians-West (WestRussian) @ 1.81 with such high percentages/low distances. I wasn't sure if it indicated ancestry from that region or was almost random (as I have no known ancestry from that area).