PDA

View Full Version : Using Faster Mutating Markers to determine connections within surname clusters



RobertCasey
04-29-2016, 03:06 AM
When I attended the 2012 RootsTech, GENETREE presented an enhancement to their 46 marker YSTR test. This new offering added 19 "faster" mutating YSTRs. I was very impressed with this new offering and their presentation. Unfortunately, GENETREE (part of Sorenson) was acquired Ancestry.com and my order was refunded. This new test was designed to build accurate genetic descendant charts at the surname cluster level. It also included a very good piece of software that built trees for you - a test with tool what a concept. Of course, this was only a demo and it could have been a well selected set of data.

This approach was similar to atDNA testing. Since 19 very fast mutating markers were added, you needed to test many descendants of a common known ancestor and it required 30 to 50 tests to produce what looked like some pretty good charts. This would be a great idea for the Irwin project where NGS tests are not enhancing the branching as was expected. You could visually see that a lot of the parallel and backwards mutations being filtered out to get to the haplotype of each oldest proven ancestor. If a pattern of mutations was not apparent, you tested more male descendants where there not enough clarity until the haplotype of your oldest proven ancestor became pretty clear. This would allow genealogists who could run down a lot of male descendants from previous traditional research to reconstruct the haplotype of their ancestor that included some very fast mutating markers. Once you had five or ten well proven lines with the haplotypes well defined that were closely matching at 67 markers (today), these currently unconnected lines could be connected using the faster mutating markers that mutated a lot in just a few generations. The pitch was well done, but I never was able to test this approach with my cousins.

The reason I bring up this approach as more YSTRs appear to be required beyond just 67/111 markers and way too many NGS tests that seem to be the trend. It appears that 111 markers and NGS tests may not be enough to break through brick walls as anticipated - specially if your cluster does not have many matches due to a genetic bottleneck. Also, both NGS tests and 111 marker tests are pretty expensive and testing 19 fast mutating YSTR markers was at a very reasonable cost. But like atDNA tests, you need multiple tests of the same proven line to determine the haplotype of your oldest proven ancestor when there were fast mutating markers included. The downside is that many more testers of the same proven descendants have to be tested - but this is somewhat encouraged already with minimal information gained. We all hoped for at least two or three YSTR based branches for each cluster. We also seem to be ignoring all the 400 YSTRs from all the NGS tests, which could be used to determine the next set of YSTR markers to include.

Here is the details of this test (from their literature in comparison with FTDNA tests):

37 Starting with the 37 marker FTDNA test
-5 Missing from FTDNA 37 marker test
+2 Moved down from 67 marker test
+8 Moved down from 111 marker test
+1 New marker not included in 111 marker test
+3 Sales pitch inflating markers by including 19b, 464e and 464f
46 Base test

The 19 fast mutating markers were:
DYF387a, DYF387b, DYF399a, DYF399b, DYF399c, DYF403Aa, DYF403Ab,
DYF403Ba, DYF403Bb, DYF404a, DYF404b, DYS518, DYS526B, DYS547,
DYS570, DYS576, DYS612, DYS626, DYS627.

It looks like 570 and 576 are found in the FTDNA 37 marker test, so only
17 new YSTRs were really added. I think in today's variation would be
probably need 20 to 25 faster markers to complement 67 markers.

Does anyone know the relative mutation rates of these 17 "faster" mutating
markers ? Also, are there another five or ten good candidates ? The NGS read
length should not become a factor, so marker selected should be only those
that Big Y can not easily determined.

If YSEQ could add these 25 faster markers, this methodology could be
tested with several surname clusters as another testing scenario.
I am not saying this methodology is superior - but it seems a worthy
testing strategy for those really want to connect unrelated lines after
surname creation time frame (usually 1,000 years). It appears that we
may have to pay for dozens of NGS tests just to solve the brick wall of
each surname cluster which just seems too expensive of a methodology.

We still need to order 10X more NGS tests, along with a lot of panel
tests, a mixture of 10 % high resolution NGS tests thrown in as well
as testing of private YSNPs individually which can be more economical
than just more NGS tests alone. I think these fast mutating markers
would be better than adding a mixture that includes so many average
and slow markers. For my analysis, 111 markers can help but just
does not seem cost effective - even over NGS higher costs.

Yet another twist is to order another 111 slow and near slow markers
- but like NGS tests, that would be at a higher costs that would
be approaching NGS tests which would be better. So FTDNA would
be right in not creating a 222 marker test at $500+.

Like atDNA testing, faster markers only have any accuracy at the under
1,000 years ago range. Attempting to use this information before
that time frame would be worthless as building YSTR only haplotrees.
However, YSTRs with around 20 to 25 % testing of YSNPs where
20 or 30 branches are available, reasonably accurrate haplotrees
can be generated.