PDA

View Full Version : Calculating TMRCA Using SNPs



JMcB
12-12-2016, 04:47 PM
Hello Everyone,

I was wondering if someone with more experience would like to give me their opinion on a question I've been trying to resolve. The question is this: how do I determine what my TMRCA is, in light of my recent BiGY results. Or do I need to, considering the information I already have.


Judging from the results I've gotten, it appears that I have split up yFull's branch of
I-L234.


L-234 Y7200 * Y3648 * Y3655 * + 11 SNPs formed 2700 ybp, TMRCA 1150 ybp.

https://yfull.com/tree/I-Y3649/


Here are their current defining SNPs for L234


13038



Of those fourteen defining SNPs I am positive or presumed positive for nine of them. While being negative for L234, Y7200, Y3654, Y3650, & Y3648.


13039



So apparently FTDNA has split me off from L234 and given most of it's defining SNPs to my branch of I-Y7623. It has also put L234 just below F3527 and left it with only four total SNPs. Placing the last of those fourteen (Y7200) just below it's new placement.


So my question is this; how would I recalculate my TMRCA in light of the above changes, using yFulls formula of (# of SNPs x 144.41 + 60)? Or does yFull's L234 estimate of 1150 ybp still apply to me in spite of the divvying up of all those SNPs?


Anyway, thank you to anyone who would like to weigh in. I appreciate it.

Regards - jmcb


P.S. I have already ordered my Bam file and will be sending it off as soon as I can.

Afshar
12-12-2016, 07:05 PM
Yfull will screen and probably find a lot more snps than ftdna, so you are never sure if you dont submit your bam to yfull first.

JMcB
12-13-2016, 01:02 AM
Yfull will screen and probably find a lot more snps than ftdna, so you are never sure if you dont submit your bam to yfull first.

Hello Afshar

As I said, I'm currently waiting for my Bam file to be generated and soon thereafter I will be sending it in to yFull. However, in the interim I would like to see if I can get an idea of when the TMRCA is. Having known myself for 63 years, I'm certainly not expecting certitude. ;-)

Dave-V
12-14-2016, 12:15 AM
I can diagram this better than I can explain it, but I'll take a shot in words at least.

You can't apply the YFull formula (# corrected SNPs x 144.41 + 60) to your full line through present day yet because for their formula you need two pieces of data that will come out of your YFull BAM analysis - the number of private SNPs in your BAM file that are also in YFull's combBED region, and the amount of the combBED region that your Big Y has covered. If you REALLY want to put in some work you can derive the first from FTDNA-reported data, but the second is harder to get and you might as well wait for the YFull report.

However you can estimate from what you have already where you branch off the 14-SNP block that YFull currently calls I-L234 and which by their estimate starts 2700 ybp and ends 1150 ybp. That block has 14 SNPs of which 12 are in the combBED region (the two that aren't are Y7200 and Y3655 as it turns out). Note that the only reason to use combBED SNPs is to align with YFull's age estimates and it doesn't mean that a non-combBED region SNP is less reliable etc.

Of those 12, you are positive for 8 of them (using your data reported here) and negative for 4 of them. That means your branch split out of that block after 2/3rds of it or about 1700 ybp (2700 - (2700-1150)*2/3 = 1666.7). After that the line of those others in YFull under "I-L234" continued to mutate L234, Y7200, Y3648, Y3650, and Y3654 ending about 1150 ybp but since that's a separate line now from you that 1150 ybp figure doesn't mean anything to your line.

That 1700 ybp estimate is merely a rough stab based on the very little data that YFull has collected so far for these branches, mind you. Your own BAM file analysis will refine that further and whether it pushes it earlier or later depends on whether you have more private SNPs or fewer than the other YFull kits under Y3649.

JMcB
12-14-2016, 01:26 AM
I can diagram this better than I can explain it, but I'll take a shot in words at least.

You can't apply the YFull formula (# corrected SNPs x 144.41 + 60) to your full line through present day yet because for their formula you need two pieces of data that will come out of your YFull BAM analysis - the number of private SNPs in your BAM file that are also in YFull's combBED region, and the amount of the combBED region that your Big Y has covered. If you REALLY want to put in some work you can derive the first from FTDNA-reported data, but the second is harder to get and you might as well wait for the YFull report.

However you can estimate from what you have already where you branch off the 14-SNP block that YFull currently calls I-L234 and which by their estimate starts 2700 ybp and ends 1150 ybp. That block has 14 SNPs of which 12 are in the combBED region (the two that aren't are Y7200 and Y3655 as it turns out). Note that the only reason to use combBED SNPs is to align with YFull's age estimates and it doesn't mean that a non-combBED region SNP is less reliable etc.

Of those 12, you are positive for 8 of them (using your data reported here) and negative for 4 of them. That means your branch split out of that block after 2/3rds of it or about 1700 ybp (2700 - (2700-1150)*2/3 = 1666.7). After that the line of those others in YFull under "I-L234" continued to mutate L234, Y7200, Y3648, Y3650, and Y3654 ending about 1150 ybp but since that's a separate line now from you that 1150 ybp figure doesn't mean anything to your line.

That 1700 ybp estimate is merely a rough stab based on the very little data that YFull has collected so far for these branches, mind you. Your own BAM file analysis will refine that further and whether it pushes it earlier or later depends on whether you have more private SNPs or fewer than the other YFull kits under Y3649.


Thank you Dave - V,

You explained that very nicely and I appreciate your efforts. Coincidentally, one of my own calculations came to that same conclusion. Unfortunately, I had four different conclusions and wasn't really sure which one was on the right track.

While I think I'll leave the heavy lifting to yFull from now on, now that you mention it, I do apparently have 15 private SNPs which are bound to alter the final results. Nevertheless, I do appreciate having a provisional number to go on.

Thanks again!

Regards - jmcb

Dave-V
12-14-2016, 02:23 AM
...now that you mention it, I do apparently have 15 private SNPs...

With those we can take a run at the YFull calculations if you'd like using an estimate of combBed coverage. I have a spreadsheet at https://www.dropbox.com/s/zm473kfyxdo5ufn/YFull-based%20SNP%20Block%20Age%20Estimator.xlsx?dl=0 that calculates how many of them are in the combBed (this has been posted before and there are a few versions around). Or if you want to PM me the positions I can run it. And assuming about an average 85-90% coverage gives us a good basis for running the calculation.

The other way to look at it is the two YFull kits already under L234 have 10 and 12 SNPs each counting their private ones and the 4 you're negative for. You're in the same ballpark; assuming a few of your 15 private SNPs are not in the combBed, your results might push the 1700 ybp estimate a bit earlier but probably not much if at all.

JMcB
12-14-2016, 05:44 PM
Hello Dave - V

Thank you for the offer but I just sent my Bam file to YFull and I'm content to wait it out at this point. Plus, I'd have to double check on what the SNPs are and I don't want to hassle my administrator at this time, when a whole new batch of BigY results are coming in.

Nevertheless, thanks again!

JMcB
02-18-2017, 01:51 AM
I can diagram this better than I can explain it, but I'll take a shot in words at least ....

However you can estimate from what you have already where you branch off the 14-SNP block that YFull currently calls I-L234 and which by their estimate starts 2700 ybp and ends 1150 ybp. That block has 14 SNPs of which 12 are in the combBED region (the two that aren't are Y7200 and Y3655 as it turns out). Note that the only reason to use combBED SNPs is to align with YFull's age estimates and it doesn't mean that a non-combBED region SNP is less reliable etc.

Of those 12, you are positive for 8 of them (using your data reported here) and negative for 4 of them. That means your branch split out of that block after 2/3rds of it or about 1700 ybp (2700 - (2700-1150)*2/3 = 1666.7). After that the line of those others in YFull under "I-L234" continued to mutate L234, Y7200, Y3648, Y3650, and Y3654 ending about 1150 ybp but since that's a separate line now from you that 1150 ybp figure doesn't mean anything to your line.

That 1700 ybp estimate is merely a rough stab based on the very little data that YFull has collected so far for these branches, mind you. Your own BAM file analysis will refine that further and whether it pushes it earlier or later depends on whether you have more private SNPs or fewer than the other YFull kits under Y3649.

Hello David-V,

Although, it's been a while I thought you might like to know what finally happened.

YFull did split me out of L234 and assigned me a new branch (Y3647) while giving me an Age Estimation of 1879 YBP. So despite lacking all of the data, your estimate was very good. My Z140 administrator who has access to an exact match of mine that YFull is unaware of, believes that our TMRCA dates to around 1500 YBP. So your estimate just about splits the difference.

For your possible interest here is how the two companies broke things down.

14116 14117

Thank you for your help, I really appreciate it.