Total versus Largest Filter default

Currently the default is the total segments method. I’d like to suggest it be Largest Segment Method since it makes cleaner graphs on average and is what most users doing segment matching prefer. However, just leaving it in the state last selected by the user might be better then either way the user won/t get a bad surprise not checking the state of the filter.

I’ve added this to the roadmap here:

1 Like

Better yet, please add both options: Total AND Largest Segment filter at the same time.

Additionally, please add both Total AND Largest Segment to EXCLUDE filter too.
Let’s say I want to exclude my maternal matches, however, in endogamy, many paternal matches also matches mother’s side as 2nd order. So I would like to exclude only 2nd order paternal matches to my mother’s side and keep only 1st order matches.
I suggested this earlier Filter Triangulation on Total AND Largest cM


If I add separate min and max cM range sliders for both Total and Largest, that should handle it, right? I don’t understand the need for separate filters on include and exclude. Isn’t it the same thing? Wouldn’t the range sliders be sufficient to specify whether to keep a kit: if the total cM match is within the total cM range specified, and the max cM is within the max cM range specified?

Or are you saying that if a kit matches an exclude filter, that you want to keep the kit if the cM match is high enough, even though it matches the exclude filter?

True Independent Dual range would be awesome for people with >20 k kits . I would for example compare the bottom 18k with the top 2k

However the two methods are different (1) setting range by Total of cm for match vs (2) setting range for smallest and largest segment size. I think the controls should be labeled “Total cM” and “Segment size” to be unambiguous.

I was arguing for a user set default because I always have to change it to Largest. :wink:

But The two could work together as radio buttons, eg “show only matches with more than 60 cM AND only segment > 6cM”.

Actually the option of dual ranges (see note) would be more use to me than being able to both screen total cM AND segment size range since I am only interested in seeing segments that are big enough to play the odds. Dual range would let different parts of the input file be sampled to get around the 20k limit

The ratio of total cM to total segments would be an interesting clue to endogamy btw but the reliability of individual segments drops off pretty steeply around 8 cM, so facilitating acquisition and testing of stuff below 6 would be a nuisance not a feature IMO. I do not want to see segments below 6 cM on the mini chromosome map in the detail views

Note:??I would like to see the segment count reported in the match detail panel as well but its meaning varies a bit by platform. Seeing a match sharing multiple segments alerts you to the possibility of sharing more than one line of descent in DNA terms. Going back to the List View to see segment count is very inefficient

Ancestry reports total cm and how many segments over 6 cm are shared. In platforms with actual data it is how many segments in the qualifying range are shared so it would have to be calculated ore fixed to a constant equivalent to AncestryDNA

Dallan, I’m saying the latter.

I think the problem is that currently, your filter only works on Total (or largest) cM between a Reference kit and matches.
When you specify INCLUDE and/or EXCLUDE kits, it is only a True or False logic that does not work in endogamy.
I would like to add Total/Larges cM filter between (Reference AND Include/Exclude kits) and matches.

For example, I would like to specify “show all kits that match ((Reference kit) AND (Include kits) over (Total 60 cM AND Largest 18 cM)) AND NOT (matching Exclude kits over Total 20 cM)”.

It would be somewhat similar to the GEDmatch “People who match both kits” with more kits option + an Exclusion logic.

Jenny, does this make sense?

It certainly extends the reach of logic. In a similar vein if I want to exclude a close cousin who by definition shares several lines, I lose every one connected in all lines in ICW matches when I am after a subset of more distant
matches within the line. In triangulation it is easier to see multiple line memberships by in ICW there is a lot of noise. One reason may be the more you share the greater the odds people who share segments on one line with you will have icw ties to people
you match on other lines because they are in fact related on that other line as well but share no dna with you.