SideView

[peter]

I have 6 DNA extended family, ranging from 132cM down to 105cM. All but 1 are in my tree and so I know which of my parents they belong to but Ancestry is suggesting the opposite on all 5!

Perhaps Stuart is right, you got Parent 1 and Parent 2 round the wrong way when Sideview for Ethnicity launched earlier in the year? I had a good idea which way round mine were, but I wasn't absolutely certain, so did nothing (as I wasn't sure whether I could change my mind later).

There is surely no way that Ancestry could have got it that badly wrong? See the statistics in #75.

 

[Helen7]

I did see a post elsewhere from someone who said they'd been told by Ancestry support that this new feature would have been rolled out to everyone by the end of Thursday (the one just gone). However, here's no sign of it in our account as yet, and since the Chromosome Painter has yet to show there, I can't say I feel that optimistic about it showing up any time soon. Hopefully I will be proved wrong!

I still don't have this feature in my account, though my husband does - frustratingly, as I'm the one paying the subscription, he just has a free account. On the other hand, I have had the Chromosome Painter for months, but as Peter says, that's not very useful.

Whilst I can log in to my husband's account and see the 'by parent' DNA match lists for both of us, for mine I don't get any 'i' labels, which makes sense as it's not my account. However, I can check the accuracy of the assignments and I haven't found any that are obviously incorrect so far. Then again, neither my husband nor I have any 1st cousin matches, and the maximum number of segments we have for any match (apart from our son) is 9, so less scope for incorrect assignment.

 

[peter]

I don't get any 'i' labels, which makes sense as it's not my account.

I'm guessing you haven't made your husband Manager of your DNA test? You won't want to do this permanently, but it might help in the short term.

 

[Helen7]

I'm guessing you haven't made your husband Manager of your DNA test? You won't want to do this permanently, but it might help in the short term.

Yes, you're right I haven't made my husband Manager of my DNA test. I don't want to do that as I want to make sure I receive all messages from DNA matches (his and mine - I am Manager of his test). Also, viewing matches through his (guest) account means I can only see partial trees. I just make a note of the ones I want to investigate then switch over to my account to look at the details.

One thing I have noticed is that in several cases of distant cousin matches, Ancestry has indicated a 'Common Ancestor' on one side of the tree, but the Parent 1/2 assignment says the match is on the other side. I suppose that isn't so strange as the 'Common Ancestor' allocation uses information from other people's trees (which is often contradictory and unconvincing) while the 'Parent' assignment takes no notice of trees but just uses DNA segment matches. I imagine the latter is more reliable, so I'm ignoring these 'Common Ancestor' hints and looking for a different connection on the other side of the tree. Does that seem a sensible approach?

 

[peter]

One thing I have noticed is that in several cases of distant cousin matches, Ancestry has indicated a 'Common Ancestor' on one side of the tree, but the Parent 1/2 assignment says the match is on the other side. I suppose that isn't so strange as the 'Common Ancestor' allocation uses information from other people's trees (which is often contradictory and unconvincing) while the 'Parent' assignment takes no notice of trees but just uses DNA segment matches. I imagine the latter is more reliable, so I'm ignoring these 'Common Ancestor' hints and looking for a different connection on the other side of the tree. Does that seem a sensible approach?

In my experience it's very rare for a 'Common Ancestor' to be wrong, but of course we're all related to everyone else in multiple ways, and the route that the 'Common Ancestor' analysis follows is simply the one that is supported by Ancestry trees,

However I should say that so far I haven't notice an examples where the Common Ancestor is at odds with SideView.

 

[Kane133]

I did an analysis of parental assignments using their SideView technology for all of Ancestry's Common ancestor hints in the DNA matches for my brother and I. Looks good to me with correctly identified common ancestors in the 8-19 cM range for which 82% were assigned a parent by SideView and were given the correct in the assignment 93-96% of the time.

 

[Kane133]

Quick decipher for the paternity 'i' label that shows (when there is a conflict between your assignment and Ancestry's SideView analysis) in the DNA matches screen:
A grey/black 'i' = unassigned by SideView but assigned by you using the "Edit Relationship" menu
A maroon/brown 'i' = assigned the opposite parent by SideView to the one you manually assigned using the "Edit Relationship" menu

Note:
1. the only cases I found of wrongly attributed parentage shared 13 cM or less;
2. all legacy matches sharing 7 cM or less are unassigned by SideView;
3. if parentage has not been manually assigned then the assignment will default to SideView meaning no 'i' as there are no conflict; and
4. also no 'i' when both manual assignment and the SideView the assignment are the same.

 

[Helen7]

4. also no 'i' when both manual assignment and the SideView the assignment are the same.

I found the same as you for points 1-3. But I still see an 'i' when my manual assignment agrees with SideView - with an invitation to change Parent 1 to Paternal side or Parent 2 to Maternal side.

I haven't found any where my manual assignment doesn't agree with SideView, so no maroon/brown 'i'.

 

[Helen7]

I did an analysis of parental assignments using their SideView technology for all of Ancestry's Common ancestor hints in the DNA matches for my brother and I. Looks good to me with correctly identified common ancestors in the 8-19 cM range for which 82% were assigned a parent by SideView and were given the correct in the assignment 93-96% of the time.

Thanks for posting this data, very interesting. It looks like my 'Common Ancestor' hints which SideView assigned to the opposite parent would fall in the margin for error in the 8-19 cM range. I found 4 of these out of a total of 60 in that range, with 3 of them being closely related to each other (mother, son and nephew) and traced back to a 'Potential Common Ancestor' that I doubt is correct. The other 'mismatched' Common Ancestor looks more plausible and I'll need to investigate further (if correct it might help me break down a brick wall!)

 

[Pauline]

A grey/black 'i' = unassigned by SideView but assigned by you using the "Edit Relationship" menu
A maroon/brown 'i' = assigned the opposite parent by SideView to the one you manually assigned using the "Edit Relationship" menu

I'm not seeing any colour differences but that may be because I haven't found the right combination of SideView assigned and manually assigned matches - ot it could be my eyes!

Where I am seeing a difference is in the wording when I hover on the 'i' - for a match where we seem to be in agreement I get "This match is on your Parent 1's side. Do you want to change Parent 1 to Maternal side? This will label all Parent 1 matches Maternal side and Parent 2 matches Paternal side", while for a few other matches I get "This is a label you selected for this match" which seems to be where SideView differs.

Looking at Common Ancestor matches (in my sister's list) there are currently 10 where I know the connection given is wrong - or at best, highly suspect. Of these 6 are given as unassigned, 3 are marked with the parent indicated by ThruLines and the other with the opposite parent.

 

[Helen7]

Where I am seeing a difference is in the wording when I hover on the 'i' - for a match where we seem to be in agreement I get "This match is on your Parent 1's side. Do you want to change Parent 1 to Maternal side? This will label all Parent 1 matches Maternal side and Parent 2 matches Paternal side", while for a few other matches I get "This is a label you selected for this match" which seems to be where SideView differs.

Yes, I see exactly these messages when logged into my husband's account and viewing his matches. The ones where I see "This is a label you selected for this match" are low cM 'legacy' matches of 6-7 cM where I'd previously allocated maternal or paternal side and which, as I understand from Kane133, are unassigned by SideView.

 

[PhilGee]

Perhaps Stuart is right, you got Parent 1 and Parent 2 round the wrong way when Sideview for Ethnicity launched earlier in the year? I had a good idea which way round mine were, but I wasn't absolutely certain, so did nothing (as I wasn't sure whether I could change my mind later).

There is surely no way that Ancestry could have got it that badly wrong? See the statistics in #75.

I'm not convinced of that - Selecting "Wales" in the chromosome painter gives me:
parent 2
chr 1, 4, 11, 12, 17, 19, 22 as all Welsh

and with one additional "region"
chr 2 English & Welsh; chr 3 Sweden/Denmark & Welsh

parent 1
chr 21, 22 all Welsh

The parent 1 "all Welsh"results were quite a surprise as I have no documented Welsh ancestry on my father's side and the nearest I get is a north Somerset "brick wall" around 1800 in my paternal line. However, it could explain why a paternal third cousin (24cM 2 segments) is allocated to my mother's side by Ancestry - though the connection is through my father's maternal line and they are west Wiltshire back to 1750's. That is the only conflict, though there are a number of >20cM "unassigned" matches that I have defined.

 

[Tim]

Perhaps Stuart is right, you got Parent 1 and Parent 2 round the wrong way when Sideview for Ethnicity launched earlier in the year? I had a good idea which way round mine were, but I wasn't absolutely certain, so did nothing (as I wasn't sure whether I could change my mind later).

There is surely no way that Ancestry could have got it that badly wrong? See the statistics in #75.

I've never used the Ethnicity stuff. But when I set them it seems to have cleared up the issues.

 

[Pauline]

The ones where I see "This is a label you selected for this match" are low cM 'legacy' matches of 6-7 cM where I'd previously allocated maternal or paternal side

The match I copied that text from is actually my sister's closest (not counting me) so I was guessing SideView had assigned it as 'Both sides', but I haven't checked this. I don't know how quickly it updates if I remove the relationship I'd added.

 

[peter]

's

I'm not seeing any colour differences but that may be because I haven't found the right combination of SideView assigned and manually assigned matches

Same here.

Where I am seeing a difference is in the wording when I hover on the 'i' - for a match where we seem to be in agreement I get "This match is on your Parent 1's side. Do you want to change Parent 1 to Maternal side? This will label all Parent 1 matches Maternal side and Parent 2 matches Paternal side", while for a few other matches I get "This is a label you selected for this match" which seems to be where SideView differs.

Same here. Typically close matches with lots of segments, so SideView will have made it 'Both sides' or 'Unassigned'.

The parent 1 "all Welsh"results were quite a surprise as I have no documented Welsh ancestry on my father's side and the nearest I get is a north Somerset "brick wall" around 1800 in my paternal line. However, it could explain why a paternal third cousin (24cM 2 segments) is allocated to my mother's side by Ancestry - though the connection is through my father's maternal line and they are west Wiltshire back to 1750's. That is the only conflict, though there are a number of >20cM "unassigned" matches that I have defined.

According to your post, chromosome 22 is shown as all Welsh on both sides, so for that chromosome at least, it's nothing to do with SideView. Ethnicity estimates will always be flaky because they're aiming at a moving target - people moved around. If you look at Ancestry's maps of the different regions you'll see that they overlap, and that some encompass seas - until the railways arrived seas were easier to cross than dry land.

Something else to bear in mind is that chromosomes 21 and 22 are the smallest chromosomes.

 

[Mitch_in_Notts]

.

One thing I have noticed is that in several cases of distant cousin matches, Ancestry has indicated a 'Common Ancestor' on one side of the tree, but the Parent 1/2 assignment says the match is on the other side. I suppose that isn't so strange as the 'Common Ancestor' allocation uses information from other people's trees (which is often contradictory and unconvincing) while the 'Parent' assignment takes no notice of trees but just uses DNA segment matches. I imagine the latter is more reliable, so I'm ignoring these 'Common Ancestor' hints and looking for a different connection on the other side of the tree. Does that seem a sensible approach?

I was also going through my Dads 'Common Ancestor' matches last night. He is 50% Scottish from his Paternal side. A relatively new match has come through with a Scottish Common Ancestor with a convincing trail in Clackmannanshire which follows with his Paternal side. The side view shows a Maternal match! He is just 5% Scottish on his Maternal side and I can't find any Scottish ancestry for his Mother (it is all Nottinghamshire and south). So at present I would say side view is wrong and I have overruled it.

 

[Kane133]

My Irish 3rd cousin once removed is 100% Irish according to Ancestry’s ethnicity estimate. However, in spite of her Maternal and Paternal ethnicity being the same and 3 out of 4 of her ‘common communities’ being the same, SideView technology was certainly very good at correctly assigning her matches as Maternal and Paternal.

Maternal
8,826 matches (4 close matches 225-764cM)
Common communities:
Connacht, Ireland
New Jersey & Eastern Pennsylvania Settlers
Munster, Ireland
Central Ireland

Paternal
8,651 matches (18 close matches 209-925cM)
Common communities:
Connacht, Ireland
Donegal, Ireland
New Jersey & Eastern Pennsylvania Settlers
Munster, Ireland

Both Sides
7 matches (9-32cM and no ‘common ancestor’ suggestions)

Unassigned
1,869 matches (8-193cM including 8 ‘common ancestor’ suggestions)

For the FAQ ... My matches aren’t split up. Why not? Ancestry say “To split up your matches, we need to find DNA segments you share with matches that cover your whole genome. If your matches aren’t split up, we weren't able to find enough shared DNA segments for you—yet. But that may change as more of your relatives take AncestryDNA tests. The more matches (especially close family matches) you have, the more likely it is that we’ll be able to split up your matches.”

So without close family matches (say 2nd cousins and closer) the Ancestry SideView algorithm will have difficulty assigning paternal or maternal.

 

[Helen7]

For the FAQ ... My matches aren’t split up. Why not? Ancestry say “To split up your matches, we need to find DNA segments you share with matches that cover your whole genome. If your matches aren’t split up, we weren't able to find enough shared DNA segments for you—yet. But that may change as more of your relatives take AncestryDNA tests. The more matches (especially close family matches) you have, the more likely it is that we’ll be able to split up your matches.”

This doesn't explain why I can't see my matches split up in my account, but I can in my husband's account (he is a 'viewer' of my matches). Despite me having only one 2nd cousin match on my maternal side and no closer than 3rd cousin matches on my paternal side, Ancestry has done a pretty good job of splitting up my matches, but I have to pretend to be my husband to see them.

Clearly, the beta SideView hasn't been rolled out to me (and Pauline) yet.

 

[peter]

For the FAQ ... My matches aren’t split up. Why not? Ancestry say “To split up your matches, we need to find DNA segments you share with matches that cover your whole genome. If your matches aren’t split up, we weren't able to find enough shared DNA segments for you—yet. But that may change as more of your relatives take AncestryDNA tests. The more matches (especially close family matches) you have, the more likely it is that we’ll be able to split up your matches.”

So without close family matches (say 2nd cousins and closer) the Ancestry SideView algorithm will have difficulty assigning paternal or maternal.

I wouldn't interpret Ancestry's explanation in quite that way. Close matches share more segments, so they contribute more to the pool of information, but I don't think they are essential to the process.

 

[Kane133]

I wouldn't interpret Ancestry's explanation in quite that way. Close matches share more segments, so they contribute more to the pool of information, but I don't think they are essential to the process.

Beg to differ Peter. Ancestry use 'phasing' to separate and identify what is paternal vs maternal DNA. The process is probably similar to how a "Lazarus" kit is built on GEDmatch for a parent who has passed. The more relatives and the closer the relatives to the 'missing' parent the better the chances of identifying DNA segments belonging to that parent/person (on GEDmatch at least). Maybe not 'essential' as Ancestry will still do it's best, but more close relatives for Ancestry to use the better the result and the fewer the unassigned matches.