* Enhancement to the Shared Autosomal DNA Diagram

For existing requests please see The Wish List
Post Reply
User avatar
ColeValleyGirl
Megastar
Posts: 3686
Joined: 28 Dec 2005 22:02
Family Historian: V7
Location: Cirencester, Gloucestershire
Contact:

Enhancement to the Shared Autosomal DNA Diagram

Post by ColeValleyGirl » 23 Jan 2022 12:44

The Shared Autosomal DNA diagram ('out of the box') shows the 'predicted' amount of Shared DNA with the Root as a %; however, anyone who has done much work with DNA will know this is a gross approximation.

Would it be possible to indicate as well the likely range of shared DNA, such as is shown in The Shared cM Project 4.0 tool v4?

(Yes, I know this would be pretty difficult, but I can only ask).

User avatar
jimlad68
Megastar
Posts: 796
Joined: 18 May 2014 21:01
Family Historian: V7
Location: Sheffield, Yorkshire, UK (but from Lancashire)
Contact:

Re: Enhancement to the Shared Autosomal DNA Diagram

Post by jimlad68 » 24 Jan 2022 15:28

I like the DNA painter table, as it seems a good empirical calculation of probable ranges for everyday use. I'm sure there will be a method to calculate these "probable ranges" but that is all they are. If you look at a simple case of siblings, theoretically they could be identical (not twins) or have no shared DNA, whilst practically probably impossible, there must be cases that veer towards that.
Jim Orrell - researching: see - but probably out of date https://gw.geneanet.org/jimlad68

User avatar
ColeValleyGirl
Megastar
Posts: 3686
Joined: 28 Dec 2005 22:02
Family Historian: V7
Location: Cirencester, Gloucestershire
Contact:

Re: Enhancement to the Shared Autosomal DNA Diagram

Post by ColeValleyGirl » 24 Jan 2022 15:35

They're actually the result of statistical calculations based on real (reported) values for shared cMs versus known relationships by the Shared cM Project.

If you click on an entry on the chart you can see a view of the underlying data.

User avatar
jimlad68
Megastar
Posts: 796
Joined: 18 May 2014 21:01
Family Historian: V7
Location: Sheffield, Yorkshire, UK (but from Lancashire)
Contact:

Re: Enhancement to the Shared Autosomal DNA Diagram

Post by jimlad68 » 24 Jan 2022 16:25

ColeValleyGirl wrote:
24 Jan 2022 15:35
They're actually the result of statistical calculations based on real (reported) values
yes, I can see that, A long time since I did statistics, probabilities etc, but I would have thought that rather than (or as well as) using limited reported empirical data, it should be possible for the DNA experts to create % probability ranges. e.g. 50% chance of being in 1 range, 10% of being in another larger range. So I think you are asking for the impossible unless you also qualify the % chance.
Jim Orrell - researching: see - but probably out of date https://gw.geneanet.org/jimlad68

User avatar
davidf
Megastar
Posts: 593
Joined: 17 Jan 2009 19:14
Family Historian: V6.2
Location: UK

Re: Enhancement to the Shared Autosomal DNA Diagram

Post by davidf » 24 Jan 2022 16:33

jimlad68 wrote:
24 Jan 2022 16:25
ColeValleyGirl wrote:
24 Jan 2022 15:35
They're actually the result of statistical calculations based on real (reported) values
yes, I can see that, A long time since I did statistics, probabilities etc, but I would have thought that rather than (or as well as) using limited reported empirical data, it should be possible for the DNA experts to create % probability ranges. e.g. 50% chance of being in 1 range, 10% of being in another larger range. So I think you are asking for the impossible unless you also qualify the % chance.
Agreed, the empirical data could be distorted by pedigree collapse and non-paternity events of which people submitting data may have been unaware.

Shouldn't this follow a typical "bell-shaped curve" as the chance of you inheriting a particular bit of genetic material from your parents is a straight 50:50 toss up (sex chromosomes excepted)?

Then the usual confidence limits and standard deviation calculations apply?
David
Running FH 6.2.7. Under Wine on Linux (Lubuntu 18.04 LTS)

avatar
Mark1834
Megastar
Posts: 1443
Joined: 27 Oct 2017 19:33
Family Historian: V7
Location: South Cheshire, UK

Re: Enhancement to the Shared Autosomal DNA Diagram

Post by Mark1834 » 24 Jan 2022 16:40

I had the same thought, but it would need to account for the fact that one side of the curve is often bounded, rather than being symmetric. My ideal is probably a simple 95% confidence interval for a given relationship.

The disadvantage of all of these options is that they are likely to be misunderstood or misused by users who don't understand the underlying stats, but in general it is better to increase education and awareness than to restrict the data to the lowest common denominator.
Mark Draper

User avatar
jimlad68
Megastar
Posts: 796
Joined: 18 May 2014 21:01
Family Historian: V7
Location: Sheffield, Yorkshire, UK (but from Lancashire)
Contact:

Re: Enhancement to the Shared Autosomal DNA Diagram

Post by jimlad68 » 24 Jan 2022 16:44

davidf wrote:
24 Jan 2022 16:33
Shouldn't this follow a typical "bell-shaped curve"
Bell Curve, my favourite, it explains so much in so many areas. I suppose my "50% chance of being in 1 range, 10% etc" implies the same or similar.
Jim Orrell - researching: see - but probably out of date https://gw.geneanet.org/jimlad68

User avatar
ColeValleyGirl
Megastar
Posts: 3686
Joined: 28 Dec 2005 22:02
Family Historian: V7
Location: Cirencester, Gloucestershire
Contact:

Re: Enhancement to the Shared Autosomal DNA Diagram

Post by ColeValleyGirl » 24 Jan 2022 16:50

the chance of you inheriting a particular bit of genetic material from your parents is a straight 50:50 toss up
Correct. But because of recombination, the chance of you inheriting a particular bit of generic material from your grandparents is not exactly 25% -- it's there or thereabouts. And the more generations you go back, the more those little differences affect the possible amounts of shared DNA.

Do any of you work with DNA at all?

User avatar
jimlad68
Megastar
Posts: 796
Joined: 18 May 2014 21:01
Family Historian: V7
Location: Sheffield, Yorkshire, UK (but from Lancashire)
Contact:

Re: Enhancement to the Shared Autosomal DNA Diagram

Post by jimlad68 » 24 Jan 2022 16:52

Mark1834 wrote:
24 Jan 2022 16:40
.... it is better to increase education and awareness than to restrict the data to the lowest common denominator.
This program is often frightening as well as interesting:
More or Less on BBC Radio 4
https://www.bbc.co.uk/programmes/p02nrss1
Jim Orrell - researching: see - but probably out of date https://gw.geneanet.org/jimlad68

User avatar
jimlad68
Megastar
Posts: 796
Joined: 18 May 2014 21:01
Family Historian: V7
Location: Sheffield, Yorkshire, UK (but from Lancashire)
Contact:

Re: Enhancement to the Shared Autosomal DNA Diagram

Post by jimlad68 » 24 Jan 2022 17:05

ColeValleyGirl wrote:
24 Jan 2022 16:50

Correct. But because of recombination, the chance of you inheriting a particular bit of generic material from your grandparents is not exactly 25% -- it's there or thereabouts. And the more generations you go back, the more those little differences affect the possible amounts of shared DNA.

Do any of you work with DNA at all?
Agreed, and "the range" is the first thing that surprised me when doing my research, but that does not stop the above scenarios fitting into a Bell Curve.
Jim Orrell - researching: see - but probably out of date https://gw.geneanet.org/jimlad68

User avatar
davidf
Megastar
Posts: 593
Joined: 17 Jan 2009 19:14
Family Historian: V6.2
Location: UK

Re: Enhancement to the Shared Autosomal DNA Diagram

Post by davidf » 24 Jan 2022 17:24

Mark1834 wrote:
24 Jan 2022 16:40
...
The disadvantage of all of these options is that they are likely to be misunderstood or misused by users who don't understand the underlying stats, but in general it is better to increase education and awareness than to restrict the data to the lowest common denominator.
There is a danger that many will take these charts as absolute "proof" (or even "disproof") rather than as indications of strong likelihood.

I am told it is theoretically possible that all the genetic material I inherited from my mother came from her mother and that I have no genetic connection with my (legitimate) maternal grandfather. Extremely unlikely - but the more removed you get the greater the chance that you have no genetic material in common with an otherwise "genuine" relative.

Which does rather raise the question what is a relative? Common DNA, or being the result of a "plumbing connection", or just a shared family heritage?
David
Running FH 6.2.7. Under Wine on Linux (Lubuntu 18.04 LTS)

User avatar
ColeValleyGirl
Megastar
Posts: 3686
Joined: 28 Dec 2005 22:02
Family Historian: V7
Location: Cirencester, Gloucestershire
Contact:

Re: Enhancement to the Shared Autosomal DNA Diagram

Post by ColeValleyGirl » 24 Jan 2022 17:38

We are wandering off topic here for this forum...

avatar
Mark1834
Megastar
Posts: 1443
Joined: 27 Oct 2017 19:33
Family Historian: V7
Location: South Cheshire, UK

Re: Enhancement to the Shared Autosomal DNA Diagram

Post by Mark1834 » 24 Jan 2022 17:57

ColeValleyGirl wrote:
24 Jan 2022 16:50
Do any of you work with DNA at all?
Not quantitatively - I keep an eye on alleged matches, and there is a good spread across various lines to indicate there are no likely "irregularities" in my recent parentage (not proof, as we could both be related to the same "wrong" person...).

Going further back, you run into the issue where most of the "matches" are likely by chance, so less useful - a bit like BT telling us our minimum guaranteed download speed is 0 mb/s - not all data are useful information.... ;).
Mark Draper

Post Reply