Filtering matches - jonathanbrecher/sharedclustering GitHub Wiki

On the Ancestry web site, it's possible to filter your list of matches to show just your Mother's Side or Father's Side matches if your mother or father has also tested on Ancestry. The same can be done with Shared Clustering on the cluster level.

Also, unlike the Ancestry web site, you can use Shared Clustering to show just the matches from the other side, if only only of your parents has tested.

The key part of filtering matches is recognizing that your close relatives show vertical and horizontal "stripes" in the cluster diagram. Here is an example cluster diagram from someone whose mother has tested but whose father has not:

In this example the red line associated with the tester's mother has a gap in it, with a little piece further down and to the right than most of the maternal matches. That's ok. You should look at the portions of the diagram that are next to the maternal red line, skipping over the gaps:

If you instead want to look at the paternal matches, those are also easy to find. Those are the ones that aren't maternal matches:

To get a nice clean cluster diagram with just the paternal matches, you should copy the Test IDs for each of the paternal matches and paste just those IDs into the "Filter test IDs to" box in the Advanced Options section of the Cluster tab:

That will produce a cluster diagram that has only the matches that have the test IDs you filtered to, in this case only the paternal matches:

Of course, if you copy the test IDs for the maternal matches, you can get a cluster diagram with only the maternal matches instead:

Filtering to grandparents

The process described above for parents will also work for grandparents (and great-grandparents), if you are fortunate enough to have grandparents who have tested at the same testing site.

Finding clusters that match a tested grandparent would be exactly the same. Finding clusters for an untested grandparent is a little bit trickier. If you simply exclude clusters from your maternal grandmother, the clusters that are left might be from your maternal grandfather, or they might be from your paternal side. In a case like that, you might look at clusters that match your mother but don't match your maternal grandmother. Adding you mother as a second filter would eliminate your paternal matches and the remaining clusters in hat case would be limited to just your maternal grandfather.

Filtering to more distant matches

The same process can be used to filter your clusters to show for example only the matches that you share with a first cousin or a second cousin. But be careful!

If you have a maternal first cousin, you can be fairly confident that all matches shared with that cousin are maternal matches to you. That's good.

You can't assume the opposite, though! In the maternal/paternal example above you could assume that any match that wasn't maternal must be paternal. The same isn't true for relatives beyond your direct ancestors. If a match is not shared with your maternal first cousin, they could still be a maternal match. You have many maternal segments that were not inherited by your maternal first cousin.

If you exclude the matches shared with your maternal first cousin, you will be left with all of your paternal matches AND some of your maternal matches that happened not to be shared with your cousin. This might still be useful in some cases, as long as you don't assume that everyone "has to be" paternal. Some of them won't be.