PDA

View Full Version : Food for thought - A possible South Asian Admixture Calculator?



jatt2016
10-15-2016, 05:12 PM
Hi Folks,

What do you think of a prospective high detail admixture calculator covering South Asian ethnic groups from Pakistan, Afghanistan, India all the way to the Bangladesh and giving percentages for population segments like :

Punjabi
Sindhi
Pathan
Bengali
Sinhali
Tamil
Hamachaperdesi
Hayanvi
Gujrati Indians
Parsi
Balochi and so on....

https://s18.postimg.org/aa7gkyg3t/Capture.jpg

Do you thinks its doable and some xperts from this forum can take a shot at it? Kurd Sahab is this something that you can undertake?:)

jatt2016
10-15-2016, 06:05 PM
Something that can give a break down like below :

https://s16.postimg.org/fps2xndoz/Capture.jpg

Kurd
10-15-2016, 07:03 PM
It is not as easy as most would think. It is hard to predict what the outcome of such an experiment would be knowing what I know about how ADMIXTURE works.

The best chance for a decent calculator like this would be to use component references that are somewhat distinct from each other, making sure that there are enough samples for each of those groups to form strong clusters. For ex, Burusho, Balochi, Bengali, Palliyar, Afghan Pashtun (although I only have 6 samples), perhaps Gujrati.

I also feel that this would be a time consuming project, but I may give it a try

jatt2016
10-15-2016, 10:00 PM
It is not as easy as most would think. It is hard to predict what the outcome of such an experiment would be knowing what I know about how ADMIXTURE works.

The best chance for a decent calculator like this would be to use component references that are somewhat distinct from each other, making sure that there are enough samples for each of those groups to form strong clusters. For ex, Burusho, Balochi, Bengali, Palliyar, Afghan Pashtun (although I only have 6 samples), perhaps Gujrati.

I also feel that this would be a time consuming project, but I may give it a try

Hi Kurd,

Thank you for giving it a thought. As you pointed it out, this would not be a simple project and then again the success of such an endeavor also depends on volunteer samples from each of the ethnic groups. My knowledge about admixture calculators is limited but I am guessing that the first step would be to find volunteers who would identify themselves as belonging to one of these groups for example :

5 People identify themselves as punjabies and submit their autosomal DNA file to you
5 People identify themselves as Balochies and sumbit their autosomal DNa file to you and so on........

and once yo have their data files you do the actual work. Is this how it works? just trying to understand the schematics here !:)

If accomplished, this would be a very interesting- first of a kind calculator and would result in greater participation/interest in genelogy from the south Asian community.

Kurd
10-16-2016, 02:09 AM
Hi Kurd,

Thank you for giving it a thought. As you pointed it out, this would not be a simple project and then again the success of such an endeavor also depends on volunteer samples from each of the ethnic groups. My knowledge about admixture calculators is limited but I am guessing that the first step would be to find volunteers who would identify themselves as belonging to one of these groups for example :

5 People identify themselves as punjabies and submit their autosomal DNA file to you
5 People identify themselves as Balochies and sumbit their autosomal DNa file to you and so on........



and once yo have their data files you do the actual work. Is this how it works? just trying to understand the schematics here !:)

If accomplished, this would be a very interesting- first of a kind calculator and would result in greater participation/interest in genelogy from the south Asian community.


I have plenty of Balochi Brahui Burusho Sindhi and Bengali samples. I only have 6 Afghan Pashtun samples. I also don't have too many S Indian tribal samples with 100K+ SNP overlap with 23andMe

jatt2016
10-16-2016, 03:16 AM
I have plenty of Balochi Brahui Burusho Sindhi and Bengali samples. I only have 6 Afghan Pashtun samples. I also don't have too many S Indian tribal samples with 100K+ SNP overlap with 23andMe

I guess then all u need would be samples from Punjabies, Sinhali and some from the indian provinces.Tamil Nado and Karnataka would be South Indian tribal?

All members welcome to give their inputs and ideas.:) We are lucky to have a knowledgeable person like Kurd who has made various admixture calculators and contributed one of the best topics/threads in this forum.

ashwinb
10-17-2016, 04:14 AM
That would be very interesting, an Eastern indian component should also be there. I would really like to see if the results are straightforwadr(not fuzzy).

khanabadoshi
10-17-2016, 07:46 AM
I have plenty of Balochi Brahui Burusho Sindhi and Bengali samples. I only have 6 Afghan Pashtun samples. I also don't have too many S Indian tribal samples with 100K+ SNP overlap with 23andMe

Perhaps McNinja might be of help? He is quite the curator of South Asian samples.

ashwinb
10-17-2016, 02:02 PM
I believe McNinja had a similar calculator: https://docs.google.com/spreadsheets/d/1XGY-UIBC2GcUWGl1FizNvu9ofle4ZPHvfNmbrVYxGMo/edit#gid=1084480849

redifflal
10-17-2016, 08:29 PM
I think we need a lot more precision within the Indian subcontinent in general. Each region's caste is largely endogamous, so even if everything is coming from a largely common pool in the paleolithic era, there should be some clustering/divergence between the various endogamous groups. I have friends from Trinidad etc that are hoping that a DNA test will get them a general idea of where in India they're from or which community, and I had to let them know that I don't think anyone has that level of precision as of now. Historically Trinidad got migrants from UP/Bihar/Jharkhand area and from Tamil Nadu. Now if an Indo-Trinidadian gets his genes to show a match with people from Andhra or Orissa, that won't be really where his origins are if he's got Tamil and Bihari ancestries.

Most of us Desis here are not like the Trinidad/Guyana/Fiji etc populations, so we know where exactly we're from. IMHO DNA companies should be testing us for free because we're last living samples of endogamous populations (for now) lol. I myself am not from an endogamous marriage, but my 2 parents are. One is eastern UP Kayastha, other is West Bengal Brahmin. I put my autosomal DNA from my Genographic test in this one calculator I don't remember the name, I was expecting it to place me possibly in between the two geographical origins, instead it put me in Bijur, Karnataka. How does that work out? If it does that for me, I can only wonder how clueless it would leave someone from the colonial-era diasporas.

jatt2016
10-17-2016, 11:05 PM
I believe McNinja had a similar calculator: https://docs.google.com/spreadsheets/d/1XGY-UIBC2GcUWGl1FizNvu9ofle4ZPHvfNmbrVYxGMo/edit#gid=1084480849

Apparently, he has compiled the results for the calculators available @ Gedmatch. Most of these calculators do not capture the south Asian fragments to the details we have wished for in this thread.