Gedmatch automated PCA for South/Central Asian members

[poi]

Finally got around to publishing the gigantic South/Central Asian Gedmatch Kit Scores. It has scores for 750+ kits from all 37 calculators from Gedmatch. I have labeled/categorized each of the kits, but if you see issues, just edit and fix. For each kit's calculator score, I also have corresponding SNP count so that we can see how much coverage each kit has for each calculator.

https://docs.google.com/spreadsheets...it?usp=sharing

If you have kits for IVC and Swat ancient samples, let me know. I can run them. Actually, I can run any kits that are not already there. Just give me the following info:

kit:
region: eg SouthAsia
nationality: eg Indian
ethnicity: eg Punjabi
group:eg Jatt
subgroup: eg Sikh

The spreadsheet is currently lacking Gangetic Plains groups except a few Brahmins. We can build up on it.

[MonkeyDLuffy]

Finally got around to publishing the gigantic South/Central Asian Gedmatch Kit Scores. It has scores for 750+ kits from all 37 calculators from Gedmatch. I have labeled/categorized each of the kits, but if you see issues, just edit and fix. For each kit's calculator score, I also have corresponding SNP count so that we can see how much coverage each kit has for each calculator.

https://docs.google.com/spreadsheets...it?usp=sharing

If you have kits for IVC and Swat ancient samples, let me know. I can run them. Actually, I can run any kits that are not already there. Just give me the following info:

kit:
region: eg SouthAsia
nationality: eg Indian
ethnicity: eg Punjabi
group:eg Jatt
subgroup: eg Sikh

The spreadsheet is currently lacking Gangetic Plains groups except a few Brahmins. We can build up on it.

Quality work as usual paji, can already see outliers in numerous groups!

[poi]

Quality work as usual paji, can already see outliers in numerous groups!

While working on the latest iteration of the gedmatch kit score spreadsheet and PCA, I noticed that one of the Khatri kits is shifted very differently, sort of almost like Nepali Chettris. I noticed the original spreadsheet had the kit noted as ("Chettri?"). Is it possible that the Khatri is mislabeled or has recent admixture with Nepali groups? Royal Chettris have marital ties with Indian Maharaja families, but I thought Khatris were not part of the (modern) Indian Royalty (despite being Kshatriyas).

[MonkeyDLuffy]

While working on the latest iteration of the gedmatch kit score spreadsheet and PCA, I noticed that one of the Khatri kits is shifted very differently, sort of almost like Nepali Chettris. I noticed the original spreadsheet had the kit noted as ("Chettri?"). Is it possible that the Khatri is mislabeled or has recent admixture with Nepali groups? Royal Chettris have marital ties with Indian Maharaja families, but I thought Khatris were not part of the (modern) Indian Royalty (despite being Kshatriyas).

Most likely, during mass pop researchs, samples get mislabeled.

[poi]

Most likely, during mass pop researchs, samples get mislabeled.

Okay, I have almost exactly 800 kits now. Out of 800, 700 seem reasonable and place within other related groups/ethnicities, but around 100 probably need to re-curated, if possible.

My plan for the kits is to create a graph of relationships and add more data like available Y/Mt DNA haplogroups. Does anyone see value in doing that? It should take a few weeks to complete, but not going to do if there is no value in doing so.

[khanabadoshi]

^^ I've already complied a list of a lot of the mtDNA yDNA for known HGDP kits somewhere. I'll have to find that spreadsheet.

[MonkeyDLuffy]

^^ I've already complied a list of a lot of the mtDNA yDNA for known HGDP kits somewhere. I'll have to find that spreadsheet.

Is it possible you can add the Tarkhans and Sainis from Sapporo's list to your list as well?

[Sapporo]

@poi

Thanks for sharing. Wanted to note one thing:

I'm positive the Maan family you posted (2-3 results) as Jatt Sikh are mixed. I'm not sure if you took them from my spreadsheets or found them yourselves but they score nothing like Jatt Sikhs (who have variation but tend to follow a pretty tight cline). It's not just the South Indian percentage but the rest of their admixture is totally off the cline. I tried reaching out to them to find out more about their background but they never responded to my inquiries.

[poi]

I am working on re-curating the spreadsheet so that outliers will be marked and/or removed. Also, my naming convention is not good, so I will clean up the spreadsheet in the next day or so. And generating this in PAST is just not interactive, so Plotly is probably where it will be.

But here are 2D PCAs -- both have similar variance coverage (a little under 90%).

PCA #1 Based on all Gedmatch calculator scores for South/Central Asians(and Parsis), the pulls are familiar. Steppe and EastAsian are pulled along the first and the second quadrants. ASI and IranN are pulled along the third and the fourth quadrants.
https://i.imgur.com/EWDp6GJ.jpg

https://i.imgur.com/fWnI1A1.png

https://i.imgur.com/gJKyNQh.jpg



PCA #2 -- the pulls in this one are a bit messier, but EastAsian and Steppe are pulling towards opposite directions.
https://i.imgur.com/JNubZxz.jpg

https://i.imgur.com/to40U3q.png

https://i.imgur.com/FlymmT2.jpg

[jortita]

The Sinhala person included definitely has mixed Sri Lankan Malay ancestry and not just Sinhala ancestry.