DNA Shared Matches

[peter]

My system is closer to Helen7's but I soon ran out of colours. I find the notes just as useful as I can't possibly remember what 24 colours mean.

(The information has to be at the beginning of the note so that it's visible.)

 

[pjd]

This is probably not the correct thread but I continue to be surprised at the numbers of 100+ cM matches that so many of you have!
I have only 2 - 1 a confirmed known 2nd cousin 162 cM & the other a 2nd cousin once removed 140 cM (both on my father's side).

 

[peter]

This is probably not the correct thread but I continue to be surprised at the numbers of 100+ cM matches that so many of you have!
I have only 2 - 1 a confirmed known 2nd cousin 162 cM & the other a 2nd cousin once removed 140 cM (both on my father's side).

The average number of 2nd cousins that we have is just 28 (according to Ancestry's research) so, even including their offspring, the number of previously unknown 2nd cousins or 2nd cousins once removed that we find is likely to be in the low single figures, and could well be zero.

Exciting though it might be to find an unexpected close cousin, distant cousins are more likely to help us knock down our 'brick walls'.

 

[jorghes]

This is probably not the correct thread but I continue to be surprised at the numbers of 100+ cM matches that so many of you have!
I have only 2 - 1 a confirmed known 2nd cousin 162 cM & the other a 2nd cousin once removed 140 cM (both on my father's side).

I often find it surprising myself - I do count myself lucky that so many of my closer cousins have tested their DNA, it does help with identifying other possible cousins with matching results. My own results only have 5 100cM or closer (not including my parents, siblings, grandmother and now one of my aunts.) and I can identify most of them.

What I find more surprising is the number of the results that come from my Australian side (which is the majority you can see in those screen shots marked with the yellow dot).

 

[peter]

What I find more surprising is the number of the results that come from my Australian side (which is the majority you can see in those screen shots marked with the yellow dot).

As I mentioned in my newsletter a few months ago, LostCousins members in Australia and New Zealand are significantly more likely to have tested their DNA than members in the UK. When you also take into the fact that people in the New World are more likely to be researching their family tree, it's not surprising that we all have a disproportionate number of matches in Australia and New Zealand.

 

[jorghes]

As I mentioned in my newsletter a few months ago, LostCousins members in Australia and New Zealand are significantly more likely to have tested their DNA than members in the UK. When you also take into the fact that people in the New World are more likely to be researching their family tree, it's not surprising that we all have a disproportionate number of matches in Australia and New Zealand.

Unsurprisingly my least number of matches come from my mother's mostly/entirely(??) UK heritage, although the majority of cousins of hers that have tested are those who emigrated to Australia and Canada.

 

[Pauline]

As mentioned above, I use the colour grouping for matches that I can definitely relate to a particular grandparent, with further 'possible' categories for matches that are shared with a known cousin but can't otherwise be identified.

On the whole this seems to work well but today I was looking at a 'distant' cousin match with whom I share 19 cM across 1 segment, but can't identify our connection from their tree. Looking at the shared matches brought up an identified cousin via my paternal grandmother (24cM, 1 segment), plus eleven more I'd marked as possibly connected in that same line. So far so good.

However, half way down the list of shared matches is an identified cousin via my maternal grandmother (24cM, 2 segments) which has me somewhat flummoxed. There is no evidence either on paper or in my DNA that my parents were related.

Is this just a coincidence, or is marking the possible connection of unidentified cousins according to shared matches with identified cousins a potentially flawed approach?

I have two matches (in the 3rd cousin category) with whom I share somewhat more DNA than expected (and more than my sister shares with them), and it seems possible I may be related to them in two separate ways. Could this be what is going on here too?

 

[jorghes]

On the whole this seems to work well but today I was looking at a 'distant' cousin match with whom I share 19 cM across 1 segment, but can't identify our connection from their tree. Looking at the shared matches brought up an identified cousin via my paternal grandmother (24cM, 1 segment), plus eleven more I'd marked as possibly connected in that same line. So far so good.

However, half way down the list of shared matches is an identified cousin via my maternal grandmother (24cM, 2 segments) which has me somewhat flummoxed. There is no evidence either on paper or in my DNA that my parents were related.

Is this just a coincidence, or is marking the possible connection of unidentified cousins according to shared matches with identified cousins a potentially flawed approach?

That is a curious incident, I'll let those more knowledge than me make a comment on the likelihood of that being possible.

I have something similar, where people from what I thought was two different branches (maternal and paternal) of a particular great-grandmother, appearing on the same list of shared matches.
I have actually wondered whether or not that simply means that my undiscovered (behind the brick wall) Jewish side is actually Ashkenazi Jewish, and related in some way to my well researched Dutch Ashkenazi Jewish side. One of my distant cousins from NZ thinks that the undiscovered side (the one we share) is actually Sephardic Jewish, which to be honest I don't think is correct, but she's been searching for longer so I can't discount it (and had better contact with members of that part of the family).

 

[peter]

On the whole this seems to work well but today I was looking at a 'distant' cousin match with whom I share 19 cM across 1 segment, but can't identify our connection from their tree. Looking at the shared matches brought up an identified cousin via my paternal grandmother (24cM, 1 segment), plus eleven more I'd marked as possibly connected in that same line. So far so good.

However, half way down the list of shared matches is an identified cousin via my maternal grandmother (24cM, 2 segments) which has me somewhat flummoxed. There is no evidence either on paper or in my DNA that my parents were related.

Many single segment matches are very distant, some are false matches. It could be either.

 

[Pauline]

Many single segment matches are very distant, some are false matches. It could be either.

So how safe is it to assume that a shared match of a known cousin (or several connecting known cousins) is possibly connected to the same line (or somewhere close to it)?

 

[Helen7]

So how safe is it to assume that a shared match of a known cousin (or several connecting known cousins) is possibly connected to the same line (or somewhere close to it)?

Peter can correct me if I'm wrong, but I think there is always the chance that a shared match in the 20-30 cM range is due to a segment (or possibly more than one?) being Identical by State rather than Identical by Descent. I'm not sure how you can tell the difference, other than if you both have extensive trees and there is no apparent connection, it would indicate at best a very distant relationship (or an unknown non-paternal event somewhere?)

 

[peter]

Peter can correct me if I'm wrong, but I think there is always the chance that a shared match in the 20-30 cM range is due to a segment (or possibly more than one?) being Identical by State rather than Identical by Descent. I'm not sure how you can tell the difference, other than if you both have extensive trees and there is no apparent connection, it would indicate at best a very distant relationship (or an unknown non-paternal event somewhere?)

I suggest forgetting about Identical by State - it's past its sell-by date.

Segments can match either because they've been inherited by the same ancestor, or they can be false matches (usually where there are crossovers, but there's also a chance of a false read). Where the segments have been inherited the common ancestor could be recent or a long way back - for example if a particular segment is common within a population as a result of endogamy.

Of course, very small segments can match purely by chance, which is why FTDNA's shared cM figures need to be looked at closely.

Ancestry downgrade common segments using their TIMBER algorithm, and their phasing algorithm should reduce the number of crossovers. FTDNA don't do either.

If you have a lot of matches with cousins who share just 1 segment and they all or mostly match each other, this tends to indicate that the segment is in a pile=up region. I try not to waste time on those matches.

 

[Helen7]

Segments can match either because they've been inherited by the same ancestor, or they can be false matches (usually where there are crossovers, but there's also a chance of a false read). Where the segments have been inherited the common ancestor could be recent or a long way back - for example if a particular segment is common within a population as a result of endogamy.

Thanks for the clear explanation. I was thinking IBS referred to the endogamous population situation you mention (that's certainly how I think of it).

I have several 'clusters' of single-segment matches of the type you refer to, and I agree that in the absence of other indications that they are genuine, are generally best ignored.

 

[Pauline]

When colour grouping unidentifiable shared matches of an identified cousin as possibly connected in the same line, my thinking to date has been that they are most likely either connected in that line or false positives. It has always seemed to me pretty unlikely that a shared match of several cousins (or even of only one) identified as connecting to a particular line will turn out to be connected to me in a completely different part of my tree. But maybe that is flawed thinking?

I have only one really large cluster of (as yet) unidentified matches, which includes 67 in the “4th cousins and closer” category (around 20% of matches in that category). With most of those I share 2 DNA segments, but with a few I share 3 or only 1.

 

[peter]

I have several 'clusters' of single-segment matches of the type you refer to, and I agree that in the absence of other indications that they are genuine, are generally best ignored.

It isn't that they're not genuine, but that they're not useful - because the connection is so far back. For example, quite a few of the those who sailed on the Mayflower came from the same area as my ancestors, so I'm quite prepared to believe that the matches I have are genuine - but it's not going to help me knock down any 'brick walls', because once I go back that far I'm related to everyone in the county.

 

[jorghes]

I was digging around in more of the shared matches of some of my relatives and I'm finding more and more crossover between my Isaacs (Jewish side with a brick wall) and Collins (Dutch Ashkenazi Jewish with detailed research going back to the late 1600s), which suggest give what Peter is saying, that it's more likely for mine to be related because of the cross over, as they don't seem to be arbitrary (and we're talking four/five different cross overs), but that the link could be further back. Though to be honest, with the intermarriage inside the Ashkenazi Jewish side (including first cousins), it's not all that surprising.

I have had the first instance of a DNA result coming up on a constructed ThruLine with what I can only describe as completely wrong - though Ancestry seems to think it's possible - the shared matches with that DNA result don't much the side of the family for the constructed ThruLine!

 

[peter]

I have had the first instance of a DNA result coming up on a constructed ThruLine with what I can only describe as completely wrong - though Ancestry seems to think it's possible - the shared matches with that DNA result don't much the side of the family for the constructed ThruLine!

ThruLines don't take account of DNA - they use trees to find possible links to your DNA matches, that's all.

 

[jorghes]

ThruLines don't take account of DNA - they use trees to find possible links to your DNA matches, that's all.

Yes I know - which is why I think the issue might need to be recognised.

 

[peter]

which is why I think the issue might need to be recognised

By who?

 

[jorghes]

By who?

By anyone using ThruLines - considering people can be confused by hints, how much more by ThruLines?