Sunday, 21 January 2018

Onset of results

The BigY, and companion Y111 upgrade, results from the last sale are starting to roll in, so it was time to republish DNASurnames (both Haplogroup overviews and Lineage/result sections), but expect further updates as quite a few results have yet to arrive.

I've not kept full track of which ones have changed when but know there were updates in the Fairbairn and McAdie projects - still waiting on those for the Andrews, Bain, Dawe, Rowe, Sinton (from the previous sale, as it had to be re-run) surnames.

Friday, 12 January 2018

Kudos to MyHeritage

The MyHeritage Year in Review blog post of 26 Dec 2017 contained this statement:

You can count on MyHeritage DNA to continue to grow rapidly and improve in accuracy and features. We are working on a major revamp of the matching algorithms which will correct old problems, add a preliminary Chromosome Browser and put accuracy at an unprecedented high level, and this will be released in early January 2018.  

It is now 11 January 2018 and they have already come good on the promise to revamp the matching algorithms (but not as far as I can see, a Chromosome Browser - but see update below).

The problems I reported earlier, my own 2nd cousin not appearing in my matches, and the predicted 2nd cousin on Ancestry and FamilyTreeDNA that was missing on MyHeritage are both now fixed.
As has one I'd not reported to them and only discovered this afternoon - a match that wasn't real but showed as reasonably strong, (45cMs approximately, 5 segments) now shows as "not related", as suspected.
Also fixed, a predicted relationship of father/daughter now rather more accurately showing as siblings!

Support was on the ball too, sent me a message advising me the update had happened, and asking for my report on whether or not my issues had been solved.
Yes.

Update:
Within a hour or so of my posting the above the promised chromosome browser was also there, with an ability to download the segment details for that individual match.
Along with a message from MyHeritage pointing me to their blog post about the improvements:
https://blog.myheritage.com/2018/01/major-updates-and-improvements-to-myheritage-dna-matching/

The match lists have grown immensely to be far more in line with those from other companies, instead of rather sparse and suspect before the upgrade.

As I've popped in our local Genie society newsletter for our DNA group news:
MyHeritage is definitely now a major player in the genetic genealogy stakes, and should deserve more consideration than Ancestry given the Chromosome Browser.

My preferred strategy should I be starting out again now would be to:
Test FamilyFinder at FamilyTreeDNA (FTDNA) to get a sample on file there available for the more specialised tests FTDNA also offer that the other companies do not - and provide hassle free communication with matches.
Then transfer that FamilyFinder file to MyHeritage to fish in that pond too - as well as GEDMatch of course.

If you've already tested at MyHeritage, that test can be transferred to FTDNA for free
https://www.familytreedna.com/landing/atdna-landing.aspx

Read Roberta Estes' post on Ancestry/FTDNA transfers before making up your mind about transferring an Ancestry test to FTDNA (they can't go the other way):
https://dna-explained.com/2017/12/04/testing-strategy-should-i-test-at-ancestry-and-transfer-to-family-tree-dna/

Don't forget GEDmatch - that cousin you've been looking for could be on any of the sites, so you should be on as many as you can too.



Sunday, 10 December 2017

DNASurnames - Haplogroup tree update

With a set of new BigY results in from the last sale, revisions have been made to the Haplogroup trees on DNASurnames - mainly under Fairbairn and Dawe but one or two others.

Check out the Recent changes.

Please note the trees are by no means all haplogroups, just those relevant to people of interest.
They show a position on the tree, with the profiles for the haplogroups sometimes including links to eg yFull, the BigTree, Surname projects for more info.

"Children" shown on a profile page include either more recent SNPs down that branch of the tree, or a patriarch for a tester of interest's line.
Each branch is collapsible/expandable to aid showing the relevant bits of the tree.

For ease of navigation, you can click on a chart link on the profile page, or in the little blue pedigree symbol up against the name for an overview of "ancestors" and "children".

Have you bitten the bullet and upgraded a relevant person for your surname of interest to BigY?
The sale price until the end of December is $100 off, but FTDNA have thrown in an upgrade to yDNA111 from whatever yDNA markers you already have.
Coupon codes are available each week that may sometimes be valid for BigY in addition to this.
I may have some available for suitable candidates in the projects I admin, or am interested in.



Sunday, 15 October 2017

Yet another MyHeritage DNA mystery

This one is another in the category of what you can't see you DO miss, ie another example of a close match reported by, in this instance, FamilyTreeDNA, and GEDMatch, but not shown on MyHeritage.

Original tests both done on FamilyTreeDNA (May 2010 and Oct 2016), both uploaded to GEDMatch and both transferred to MyHeritage.

Actual relationship 2nd cousins.

FamilyTreeDNA:
Predicted relationship 2nd to 3rd cousin sharing a total of 144cMs, largest segment 19 cMs
Looking at only those segments over 7cMs this reduces to a total of 96 cMs over 7 segments  (including a 10cMs X chromosome match)




Chromosome Start Location End Location centiMorgans (cM) # of Matching SNPs
1 90373522 107499183 14.1 4500
2 50979064 71469042 18.47 5500
3 173752942 179284891 8.02 1300
5 171792682 180625733 18.64 2613
17 74799181 78639702 8.85 1023
18 4458642 11750915 17.58 2693
X 131769934 139569333 10.25 925

GEDMatch 
One to One comparison at the default settings of 500 SNPs minimum 7cM segments
Comparing Kit A and Kit B
ChrStart LocationEnd LocationCentimorgans (cM)SNPs
190,080,689107,508,90715.54,344
250,728,18571,670,51117.95,333
3173,574,411179,657,3938.61,369
5171,696,763180,623,54320.52,515
1774,702,72278,637,19810.81,013
184,326,19111,866,61328.12,716
2219,337,48523,815,8029.7836
Largest segment = 28.1 cM
Total of segments > 7 cM = 111.1 cM
7 matching segments
Estimated number of generations to MRCA = 3.5

672062 SNPs used for this comparison.


MyHeritage
This maternal cousin does show up on the MyHeritage match lists for my two maternal aunts, both 
sharing 5.5%, 395cMs total, largest 63cMs, one with 14 segments, the other with 18.
But nowhere to be found on my match list, nor I on hers.

Looking at the FTDNA Chromosome Browser picture for my 2nd cousin compared to myself and the maternal aunts clearly shows at least three segments where I match at least one of the aunts (chr. 1,2,5) at which spots those matches total about 52cMs, well above what I'd consider a match from MyHeritage


Has been logged with MyHeritage Support for their scientists to take a look

Wednesday, 20 September 2017

Plea for MyHeritage DNA

Oh I do so wish MyHeritage would hasten along their promised Chromosome Browser so that where a match actually matches can be easily seen without a plea for them to upload to GEDmatch - although the latter upload is highly recommended regardless *.

* bewitched bothered and bewildered by GEDMatch?
Check out Jim Bartlett's latest post on his segmentology blog: Getting started with GEDMatch



My second wish is that such a tool would also differentiate between
- the components of the match that are real, from comparing apples with apples, and
- those bits that are imputed by MyHeritage in order to bring the set of SNPs being compared up to the full set
which latter comparison is apples with bananas, the last word being chosen deliberately.


My assumption is that this latter imputation is responsible for the example below (logged with MyHeritage for an explanation, and the two pleas above.)

An excited match contacted me about one of my kits saying "I'm a 1st cousin 2*removed can I see your tree please" adding later that "the connection should be easy to find".

Leaving aside that the tree was connected and visible, and that the match is actually listed under the heading MyHeritage labels as:
"DNA Matches listed below are lower quality matches that should be viewed with skepticism"
what MyHeritage reported was 

ie a total of 31.8cMs over 3 segments, largest segment 15.6cM

What Gedmatch reported at the default settings for the two ancestry kits, both presumed to be recent, ie post their chip change May/Jun 2016

Comparing Kit one and Kit two
Minimum threshold size to be included in total = 500 SNPs
Mismatch-bunching Limit = 250 SNPs
Minimum segment cM to be included in total = 7.0 cM

Largest segment = 0.0 cM
Total of segments > 7 cM = 0.0 cM
(2228) No shared DNA segments found

441115 SNPs used for this comparison.


Lowering the minimum segment cM to 5cM showed two small segments:
 
Chr Start Location End Location Centimorgans (cM) SNPs
1 120,064,541 151,437,807 5.5 684
9 127,173,028 131,032,682 5.1 524
Largest segment = 5.5 cM
Total of segments > 4 cM = 10.6 cM


We cannot yet find, and do not expect to, this match on Ancestry.

Wednesday, 30 August 2017

Assorted DNASurnames updates

Several updates have occurred to the DNASurnames Haplogroup tree that shows key branches, the patriarch, and optionally, the tester, for surnames/lineages of interest to assorted projects, sometimes sub-projects of much larger ones.

The haplogroup R tree now includes in particular:
- BigY driven changes around R-S7361, including the new branch created for my Henderson family
- ditto around R-ZZ7_1 (well R-Y16467*) for the Richardsons of Morebattle subset of the Richardson-2 project
- a place holder at R-M269 for the Dawe family of Lamerton, Devon while we await BigY test results for the line

On the lineages side, the RICHARDSON pages in particular have had a revamp to include a theoretical DNA signature chart linking up the 4 yDNA STR matching lines to their parent terminal SNPs to see what the derived ySTR signature for the progenitor would be.

Another MyHeritage mystery

Previous posts have documented some of the vagaries experienced with MyHeritage's DNA matching.

This one makes me wonder how those without an ability to compare at other companies would even  realise that they may be missing a close match!

I expect that some matches may well be false positives given the imputations MyHeritage does to compare test kits from assorted companies, but I didn't expect to NOT find a match reported as predicted 2nd cousin by Ancestry, 3 generations by GEDmatch and predicted 2nd to 3rd cousin by FamilyTreeDNA.

The comparison details:

Ancestry: both tested there
Predicted relationship: 2nd Cousins
Possible range: 2nd - 3rd cousins
Confidence: Extremely High
Total shared: 208 cMs over 11 segments

FamilyTreeDNA: one tested, the other transferred in from recent-ish Ancestry test
Predicted 2nd to 3rd cousins total 202 * cMs, largest segment 57
* effectively 175cMs when < 5cM segments excluded

GEDMatch: one uploaded from FTDNA, the other from the recent-ish Ancestry test #
At the defaults for one to one comparison:
Largest segment = 56.2 cM
Total of segments > 7 cM = 212.5 cM
11 matching segments
Estimated number of generations to MRCA = 3.0

# these kits being the same used for the MyHeritage uploads.

This has been reported to MyHeritage Support for their comments and explanation.