DNA: the Next Step in Family History

Mayo Genealogy Group

2:00 p.m. Saturday 13 May 2017

National Museum of Country Life, Turlough Park, Castlebar, County Mayo

WWW version:

Facebook event:

https://www.facebook.com/events/707451316082908/

YouTube version:

As there is no 3G mobile broadband coverage in Turlough Park, I could not use my own laptop for this presentation and therefore could not record it. The Minister for museums and rural broadband has promised many times to rectify this situation at some unspecified future date.

What is Genetic Genealogy?

Genealogy is the study of family history using:

archival sources (censuses, birth/marriage/death (BMD) records, etc.); and
oral tradition (interviews, written memoirs, etc.).

Genetic Genealogy augments this by also using:

DNA (short for deoxyribonucleic acid)

samples collected using saliva or cheek swab kits
one for everyone in the Late Late Show audience on St. Patrick's Day

Genetic genealogists combine:
- online DNA data; and
- online pedigree charts
to help long-lost cousins to find each other and to find their long-forgotten ancestors.
Technology has now reached the stage that not using DNA is considered the equivalent of not using census returns.
We have a duty to our descendants:

to preserve our oral history by committing our family memoirs to writing; and
to preserve our genetic history by submitting our DNA to a DNA database.

Since the Genographic Project Team visited the National Museum of Country Life as part of The Gathering Ireland 2013:

costs have fallen dramatically;
technology has improved dramatically; and
database sizes have grown exponentially.

Where does our DNA come from?

DNA is:

made up of chromosomes and mitochondria, each consisting of molecules of four nucleotides named adenine (A), cytosine (C), guanine (G) and thymine (T)
represented by strings of the letters A, C, G and T

When a sperm fertilises an egg, each brings DNA, which is replicated in every cell of the resulting person.

	male offspring	female offspring
sperm	Y chromosome	X chromosome
	22 paternal autosomes
egg	X chromosome
	22 maternal autosomes
	mitochondria

autosome is short for autosomal chromosome.

Inheritance paths

Y chromosome: Only males have a Y chromosome.
The Y chromosome comes down the patrilineal line - from father, father's father, father's father's father, etc.
This is the same inheritance path as followed by surnames, grants of arms, peerages, etc.
X chromosome: Males have one X chromosome, females have two.
X DNA may come through any ancestral line that does not contain two consecutive males.
Blaine Bettinger's nice colour-coded blank fan-style pedigree charts show the ancestors from whom men and women can potentially inherit X-DNA.
Autosomes: Exactly 50% of autosomal DNA comes from the father and exactly 50% comes from the mother.
On average 25% comes from each grandparent, on average 12.5% comes from each greatgrandparent, and so on.; Due to recombination (see below), one might inherit, for example, 27% from the paternal grandfather and 23% from the paternal grandmother.
Siblings each inherit 50% of their parents' autosomal DNA, but not the same 50% (except for identical twins).
Mitochondria: Everyone has mitochondrial DNA.; Mitochondrial DNA comes down the matrilineal line - from mother, mother's mother, mother's mother's mother, etc.
The surname typically changes with every generation in this line.

For genetic genealogy, beginners should start with autosomal DNA, or with Y DNA for one name studies or surname projects.

How much DNA do we have?

Billions of letters:

	Male			Female
	Length	Width	Total	Length	Width	Total
Autosomal	2,881,033,286	2	5,762,066,572	2,881,033,286	2	5,762,066,572
X	155,270,560	1	155,270,560	155,270,560	2	310,541,120
Y	59,373,566	1	59,373,566			0
Mitochondrial	16,569	1	16,569	16,569	1	16,569
GRAND TOTAL	3,095,693,981		5,976,727,267	3,036,320,415		6,072,624,261

How much DNA do we observe?

The three major DNA companies (see below) sample different locations on the autosomes.
The locations sampled are about 0.02% of the total, but are those known to vary between individuals.
The vast majority of DNA is identical for all humans, and even for many of the apes.

FamilyTreeDNA (Family Finder): 682,608
AncestryDNA: 441,447 (reduced from 658,780 since May 2016)
23andMe: 577,382

The overlap between FamilyTreeDNA and the original AncestryDNA set was 652,462 (based on my personal results).

The random component of DNA inheritance

Most DNA is transcribed exactly from the relevant parent.

Two sources of randomness mean that one cannot always exactly infer the child's DNA from the parents' or vice versa:

mutations
recombination

Mutations are transcription errors at single locations, e.g. a single A in the parent may be replaced by a C in the child.

Some locations mutate very frequently (every couple of generations), and can be used to identify individuals beyond reasonable doubt, e.g. in criminal cases.

Some locations mutate less frequently (only once in many generations or once in the history of mankind), and can be used to identify closely or distantly related individuals.

Special types of mutations:

Short Tandem Repeat (STR): a string of letters consisting of the same short substring repeated several times, for example CCTGCCTGCCTGCCTGCCTGCCTGCCTG is CCTG repeated seven times; it may be repeated less or more often in other individuals.
Single Nucleotide Polymorphism (SNP): a single location where two (or occasionally more than two) different letters are observed in different individuals.

Y-DNA Mutations

The entry-level Y-DNA37 product looks at the numbers of repeats for each of 37 STR markers on the Y chromosome, e.g. Durkin/Durkan/Durcan Surname Project
Note that surname spellings also mutate, independently of DNA mutations.
Some SNPs on the Y chromosome are once-in-the-history-of-mankind events and can be used to build a Y-DNA Haplogroup Tree.
STRs can only predict Y haplogroups but a SNP product must then be purchased to confirm the Y haplogroup.
Surname-specific SNPs are now being discovered.

mtDNA Mutations

There is also an mtDNA Haplogroup Tree.

Autosomal DNA Mutations

The 0.02% or so locations observed on the autosomes are known SNPs.
All observed SNPs are still weighted equally in relationship calculations, but two individuals who share a mutation observed in only 1% of the population are likely to be far more closely related than two individuals who share a mutation observed in 50% of the population.

Recombination

The other source of randomness is recombination, which is how, e.g., the father's paternal and maternal autosomes cross over to produce the child's paternal autosomes.

Every sperm and egg is potentially unique.
Recombination of the paternal and maternal chromosomes is sometimes compared to shuffling two decks of playing cards.
Recombination rates vary markedly along the autosomes and the X chromosome.
The local recombination rate is referred to as genetic length and is measured in units called centiMorgans, which estimate the number of generations to the common ancestor of two individuals sharing an identical long run of letters.
One recombination per generation is expected in each 100 centiMorgans (cM, not cm).
The longer the centiMorgan length of two identical DNA segments, the more recently one can expect to find the common ancestor from whom they were inherited.
On average, there are 798,852 total letters per cM or 190 letters observed per cM, or one recombination or crossover per generation per 19,000 observed letters.

The double negative approach: opposite homozygous locations and half-identical regions

Current technology can observe two letters at each location, but cannot distinguish between the paternal and maternal letters.
Future technology may allow segments of one person's paternal or maternal chromosome to be compared directly with segments of another person's paternal or maternal chromosome.
In the meantime, autosomal DNA matching is based on a double negative approach:

The two of you might have inherited your DNA from a common ancestor at any location except where you definitely did not.
Example

In practice, almost all of the observed locations are bi-allelic - i.e. only two letters have been observed in the human population.
At a bi-allelic location (e.g. where only A and G are observed):

an individual with different letters from each parent (e.g. AG) is called heterozygous
an individual with the same letter from both parents (e.g. AA or GG) is called homozygous.

How can we tell whether two individuals inherited a particular location from a shared ancestor?

two individuals who are opposite homozygous at a particular location (e.g. one AA and the other GG) definitely did not inherit that location from a common ancestor.
otherwise we cannot rule out a common ancestor:

AA and AG could have the A from a common ancestor;
GG and AG could have the G from a common ancestor;
two AGs, two AAs or two GGs could have both letters from common ancestors (and may even be doubly related).

Between one opposite homozygous location and the next, the chromosomes are broken up into half-identical regions where at least 1-out-of-2 letters is common to both individuals.

Long half-identical regions could contain identical segments inherited from common paternal/paternal, paternal/maternal, maternal/paternal or maternal/maternal ancestors.
The longer a half-identical region, the more likely it is to contain an identical segment (probably with fuzzy boundaries at each end), and the higher the probability of a recent common ancestor.
To find possible relatives, concentrate on runs of 1000 or more SNPs with no opposite homozygous locations.
A half-identical region is described by:

Chromosome number (1-22 or X)
start location
end location
estimated centiMorgan length
number of SNPs observed
names of the two parties being compared.

e.g.

5 149,295,480 169,691,785 27.0 2443 John/Ger

Match list (log in as individual kit number or as project administrator managing several people's kits)
Chromosome browser (log in as individual kit number)

False positives can occur - a long half-identical region can by chance consist of short overlapping (or zig-zagging) paternal/paternal, paternal/maternal, maternal/paternal and maternal/maternal segments.

Phasing and triangulation are like opposite sides of the same coin:

If you are half-identical to two people on the same region, where they are not half-identical to each other, then one must match your paternal chromosome and the other must match your maternal chromosome:

phasing is the term used for separating the observed letters into paternal and maternal chromosomes, or separating the observed matches into paternal and maternal relatives.

If you are half-identical to two people on the same region, where they are half-identical to each other, then all three of you must have inherited DNA in that region from a single common ancestor:

triangulation is the term used for identifying three people descended from a common ancestral couple. (More information is usually required to determine whether the shared DNA came from husband or wife in that couple.)

Rule of thumb for lengths of the longest half-identical region::

30 cM+: good chance of finding the common ancestor
20 cM+: worth investigating
10 cM+: probably too distant to trace in surviving Irish genealogical records
under 10cM: possibly half-identical by chance
reduce these thresholds if there is other solid evidence of a relationship, such as common ancestral surnames or common ancestral locations
50 cM+ in extreme cases (see below) can still be shared by fourth cousins or even more distant cousins

The aggregate length of all the half-identical regions above some arbitrary threshold is used to estimate the relationship:

1cM threshold for FamilyTreeDNA (provided one segment is 7cM or more)
7cM threshold for GEDmatch

Average autosomal DNA shared by pairs of relatives:

in theory; and
in practice.

Rules of thumb:

over 100 total shared cM - closer than 4th cousin;
under 100 total shared cM - more distant than 2nd cousin.

over 100 total shared cM - use "Shared Centimorgans" to find closest relatives;
under 100 total shared cM - use "Longest Block" to find closest relatives.

3rd cousin and closer: typically more than one significant half-identical region
4th cousin and beyond: typically just one significant half-identical region

What can DNA tell us?

With one sample, you can fish for long-lost cousins in three commercial pools and one non-commercial pool (GEDmatch.com).
With two samples, you can confirm or disprove theories about relationships (e.g. Richard III).
It can potentially break through all sorts of genealogical brick walls.
For people with adoptions (or foundlings) in their family history, it may be the only option to find their genetic family; such people are disproportionately represented in all the databases, even compared to the 2% adoption rate in the past.
Those whose genetic ancestry has been concealed from them have a right, and usually a great desire, to know it.
On the other hand, if there is a family secret that you, or others in your family, would like to remain a secret, then genetic genealogy may not be for you.
Ethnicity
- sensitive to the (unspecified) reference population
- distorted by recombination (e.g. 33% rather than 25% from one grandparent)
- "genetic astrology"
- "pseudo-science at its worst, running on the unspoken but still queasy implication that race has some scientific basis" (John Grenham)
- most Irish people are within the margin of error of 100% Irish (or 100% British Isles according to one company)
- great marketing value in the U.S. melting pot
Data-mining, lies, damned lies and statistics

The big 3 DNA companies

New competitors are emerging: Living DNA, MyHeritage

AncestryDNA: Part of ancestry.com
Autosomal DNA only; Very limited analysis tools
Represents only 30 countries as of October 2016 and historically has either overcharged or not accepted non-U.S. customers; Full access requires paying ongoing annual subscription
Internal messaging system
Reached 1 million samples in July 2015; 2 million samples in June 2016; 3 million samples in January 2017; 4 million samples in April 2017; Most people use pseudonyms or initials and conceal their real surnames
My results
23andMe: Concentrates on medical aspects of DNA; Autosomal DNA plus predicted Y-DNA and mtDNA haplogroups
Overcharges non-U.S. customers; One-off payment
Optional internal messaging system
About 1 million samples; Most people anonymous; Analysis tools for non-anonymous matches
"We will soon transition your account to the new 23andMe experience" and "Preparation for the transition to the new 23andMe experience" (including doubling of prices) started in October 2015 and still ongoing as of May 2017
Results
Surname View
FamilyTreeDNA (FTDNA): Dedicated to genetic genealogy
Autosomal DNA (Family Finder) plus various Y-DNA and mtDNA products; Good analysis tools; Single worldwide price (sale for U.S. Mother's Day); One-off payment; No U.S. bias; Simple e-mail communications; 873,827 records as of 12 May 2017; Most people use real names: but married women are recommended to use maiden surnames; Projects - e.g. Clare Roots and surname projects; Time for someone to start a Mayo Roots project (not to be confused with the Mayo Surname project); You can transfer your existing National Geographic Genographic Project Y-DNA and mtDNA results to Family Tree DNA for free and upgrade to Family Finder for just USD39
My results

What do you get for your money?

An online list of DNA matches, ordered by closeness of relationship
Regular updates as the database grows
Ethnicity estimates
Raw data to use at third-party websites
Various tools for analysing matches and data
Surname projects (FTDNA only)
Other projects (FTDNA only)

The third-party sites

GEDmatch.com

Whichever of the big three companies you use, you MUST copy your results to GEDmatch.com and persuade your DNA matches to do likewise - see here.

for the vastly superior analysis tools; and
for the vastly increased chance of finding someone who will be able to confirm long-forgotten relationships

Even if the GEDmatch transfer process taxes your patience, you can also share match lists

by joining an FTDNA project; or
by sharing an AncestryDNA match list: AncestryDNA home, SETTINGS, Sharing DNA results section, INVITE OTHERS TO ACCESS DNA RESULTS, enter Email or Ancestry username, tick Guest button, SEND INVITATION

There is a trade-off between maintaining privacy and finding your long-lost cousins: you must be prepared to give up some privacy if you want to find your relatives.

DNAgedcom.com

featuring ADSA

Levels of involvement

At a minimum, please consider my advice on how to get the most out of your DNA purchase.
Others can then help you to analyse your results.
One's involvement can be at different levels:

Lists of names: A black box algorithm can be used to list the names of those in a database whose autosomal DNA is closest to yours.; You can look at your matches' own names, their ancestral surnames, ancestral placenames and family trees, if they have made these available.; Anybody can do this.
Lengths of half-identical regions: To get full value from one's investment in DNA analysis, one should move on from the purely qualitative approach of looking at names and take a more quantitative approach.; The first step is to look at the centiMorgan lengths of the regions on which one is half-identical with a potential relative.; The higher the centiMorgan numbers, the closer the relationship is likely to be.; Some basic arithmetic skills are required for this.
Locations of half-identical regions (phasing and triangulation): This may exercise your brain cells a little more than the first two approaches.
Raw data: Sooner or later, the only answer to a particular DNA puzzle will be to look at one's raw data, in the form of long sequences of pairs of As, Cs, Gs and Ts, in order to work out exactly how and why something happened.; This is for the specialist. There are many of us around, willing to help.

Example I: O'Neills of Shammer, Kilmovee

The late Des O'Neill of Swinford and Omeath provided manuscript pedigrees: version 1; version 2
It would be easy to confuse Thomas O'Neill, farmer, and Thomas O'Neill, labourer, both of Shammer.
Both Thomases had sons named Michael married to women named Mary less than 12 months apart in 1868 and 1869.
The sons had the same occupations as their respective fathers.
In the case of the farmers, the groom's residence at the time of marriage was Shammer and the wife's maiden surname was Glavey.
In the case of the labourers, the groom's residence at the time of marriage was Shamer and the wife's maiden surname was Lyons.
Only one Thomas, presumably the farmer, appears in Shammerbaun or Shammerdoo or Barcull in Griffith's Valuation (15 Nov 1856) - no. 8 in Shammerdoo.
Des O'Neill may have merged the two families.
Des shows that Mary Glavey's mother-in-law was S*bina Durcan.
Other evidence confirms that Mary Lyons's mother-in-law was Sabina (maiden surname unconfirmed).
Des shows elsewhere that one of the Thomas O'Neills was a brother of my GGgrandfather Edmund O'Neil of Barnalyra.

DNA comparisons between a GGGgranddaughter of Thomas the farmer (A115380) and GGgrandchildren (T734272, T392620 and T205074; first cousins to each other) and GGGgrandchildren (A663947 and A596212; third cousins to each other; third cousins once removed to the others) of Edmund confirm that they were closely related (Excel version):

Kit	A115380	T734272	T392620	T205074	A663947	A596212
A115380		7.1	19.2	23.0		16.0
T734272	7.1		928.1	921.0	33.1	9.8
T392620	19.2	928.1		1126.0	49.0	59.1
T205074	23.0	921.0	1126.0		16.4	109.4
A663947		33.1	49.0	16.4		65.5
A596212	16.0	9.8	59.1	109.4	65.5

Shared cM between the first cousins ranges from 921.0cM to 1126.0cM.
Shared cM between the third cousins is 65.5cM.
Shared cM between the third cousins once removed ranges from 9.8cM to 109.4cM (A596212 is actually a double third cousin once removed, sharing both Durkan and O'Neill ancestry with the three Txxxxxx kits).
Average for the three third cousin once removed relationships is 32.8cM.
Average for the three double third cousin once removed relationships is 59.4cM.
Des O'Neill's theory says Thomas's descendant is a fourth cousin once removed and fifth cousin to Edmund's descendants.
Shared cM Project found average of 20cM for 4C1R; the results in the O'Neill example are 7.1, 19.2 and 23.0 (average 16.4).
Shared cM Project found average of 17cM for 5C; the results in the O'Neill example are 0 and 16.0 (average 8.0)
The common ancestor might still be a generation further back.
The sample size of five known cousins is too small to draw definitive conclusions.

Example II: A 1938 adoption: Baby Ann

The Adoption (Information and Tracing) Bill 2016 is currently going through the Oireachtas.
It currently makes no reference to DNA or genetic genealogy.
Does our government realise that recent scientific progress has let the genie out of the bottle as regards adoption secrets?
Reuniting families separated by adoption, whether using DNA or more conventional methods, should not be attempted without the advice of an experienced genetic genealogist and/or an experienced social worker.
There will only be one chance to make the critical first contact a success.
Anthea (T575740) was the first match to contact me in November 2013.
She was adopted in 1938.
Her story has received widespread media publicity worldwide for almost 80 years (sample article 1 [with video]; sample article 2).
DNA matches point to one parent from Connemara and the other parent from the Swinford/Charlestown/Kilkelly triangle.
The first major breakthrough in 2016 involved identifying a number of known second cousins to whom she appears to also be a second cousin.
Their most recent common ancestral couple (O'Donnells of Mayo) must be two of her eight greatgrandparents.
Further leaps forward have been made in her case by obtaining DNA samples from relevant cousins of existing matches.
Potential parents have now been identified, but the tricky part is approaching potential half-siblings.

Example III: Marriage Dispensation - Stenson/Durcan

There is an intriguing parallel between DNA matching and dispensations noted in marriage registers.
Both tell us that two people have a common ancestor without telling us on which side of either family that ancestor is.
The present primitive state of DNA matching technology and the Catholic church's inability or unwillingness to produce detailed applications for dispensation from its diocesan archives are equally frustrating for genealogists.
In an ideal situation, a note of a dispensation could act as corroboration for DNA evidence and vice versa.
7 Mar 1859:
Patritius Stenson Michael Durcan In 4to et 4to
Honoria Durcan Brigida Durcan con. grad.
(last entry on right-hand page)
My greatgrandmother was Honoria's sister.
Mary's greatgrandfather was Patritius's brother.
So Mary and I are sixth cousins (biggest half-identical region 4.7cM, 253 SNPs; different AncestryDNA chips).

Example IV: Waldron Y-DNA

There are several different theories as to the origin(s) of the Waldron surname
The oldest occurence of the present English-language spelling in Ireland is in the Ulster plantation of 1609
A much older Ballyhaunis area Norman surname is now also spelled Waldron in English and de Bhaldraithe in Irish
My GGgrandfather Thomas Waldron was born in County Roscommon and joined the then Irish Constabulary in 1847
Was he related to the Waldrons of Ballyhaunis?

My Y-DNA is from haplogroup R1b
The first Ballyhaunis Waldron tested is from haplogroup R1a
So no common male line ancestry for 25,000 years or more

Example V: My Lynch ancestor was Irish - what townland was he from?

Eugene/Owen Lynch was born in "Ireland" around 1869 and died in San Francisco in 1917.
He was grandfather to seven siblings who are all in the DNA databases.
Oral tradition is that an O'Dea farm in Moveen West, County Clare, was divided between two sisters, who married Lynch and Connell respectively, around 1820.
In the Tithe Applotment Book of 17 Nov 1827 for Moyarta parish "Mich'l Lynch 7 Div" occupied a total of 15 Irish acres and two lines below him "Patt Connell 7 Div" occupied a total of 16 Irish acres, the difference being 1 acre of 4th quality land.
In Griffith's Valuation, Mary Lynch née O'Dea had the entire Lynch holding of almost 69 statute acres (no. 9a) but the Connells are omitted.
We have DNA samples from Claire (whose mother was a Lynch) and Ed (whose mother was an O'Connell) and based on this oral tradition they are fourth cousins:
```
8 8,292,285 54,555,905 50.2 10692 Claire/Ed
```
This shared segment, unusually long for fourth cousins, must have come through the two O'Dea sisters.

Claire and Ed triangulate with several of Eugene Lynch's grandchildren, e.g. Michael:

8 17,609,398 38,378,638 27.4 5563 Ed/Michael
8 17,648,866 37,812,773 26.9 5499 Michael/Claire

So, combining evidence from DNA, dates and surnames, Eugene must have been a grandson of Michael Lynch and Mary O'Dea who married around the early 1820s and whose first names are known from the death certificate of their son Daniel who emigrated to Newtown, Connecticut.
We are lucky that the shared DNA came through a female ancestor and the newly discovered relatives have inherited her husband's surname, so there is no doubt where they fit in the family tree, apart from the first name of Eugene's father.
Five Lynch siblings and one of their children were in Moveen in November 2016.
Their potential second cousin has provided a DNA sample, currently being processed.
Meanwhile other descendants of Michael Lynch and Mary O'Dea have come to light via the DNA databases.
Photo album (for Facebook friends).
"That day was the highlight of our trip to Ireland, and for me it was an unbelievable gift. I have been wondering and imagining for so long about my grandfather's (and thus my own) origins that I still quite can't believe I went there. It was a profound and moving experience for me. My sister Pat and I speak of it often ... It was magical to us. Both of my parents were from families who didn't stay connected with their homeland, and this journey has really fulfilled a need that I hadn't realized I had. I'll be forever grateful to Tom Kearney for introducing me to you, and to you for so generously sharing your expertise with us."

Conclusion: Why you should submit your DNA

The value of DNA "testing" to genealogists increases dramatically with the number of people from the relevant geographical area and relevant extended family group already in the DNA databases used.
Submitting your DNA to a database has significant positive externalities for existing and future researchers.
We need to persuade more Irish people to submit DNA samples to the databases for purely genealogical purposes.
Your descendants will be eternally grateful to you for leaving them your DNA.