PDA

View Full Version : Frustrated by FTDNA'S X-matching



firemonkey
08-05-2017, 01:20 PM
The criteria for a match seem ridiculously low especially with regards to matches to a female if you are male. Another match today with a female only to find it is 2.37 cMs.
I think 23andMe's matching criteria according to https://isogg.org/wiki/Autosomal_DNA_match_thresholds are far better.

Saetro
08-05-2017, 09:00 PM
The criteria for a match seem ridiculously low especially with regards to matches to a female if you are male. Another match today with a female only to find it is 2.37 cMs.
I think 23andMe's matching criteria according to https://isogg.org/wiki/Autosomal_DNA_match_thresholds are far better.

Agree.
Furthermore, the X match column simply says "X match": there is no statement of cM, and there is no ordering of matches by cM length.
(It is possible to sort by X match length by downloading your Chromosome Browser matches and sorting the Excel file from that.)
But once on the Chromosome Browser you can easily raise the threshold to 10cM and remove tiddlers.

geebee
08-06-2017, 05:55 AM
The criteria for a match seem ridiculously low especially with regards to matches to a female if you are male. Another match today with a female only to find it is 2.37 cMs.
I think 23andMe's matching criteria according to https://isogg.org/wiki/Autosomal_DNA_match_thresholds are far better.

In general, I'd agree. But realize that these different match thresholds can lead to some odd results. For example, my sister CJ and I share three segments on the X chromosome. My daughter Kathryn inherited this chromosome from me, since it's the only X chromosome I have. Therefore, she ought to share the same three segments with CJ that CJ shares with me, but she only shares the first and last.

The reason is the female-female threshold difference. At 23andMe, the higher threshold is not only for length, but for number of shared SNPs. The match doesn't meet the threshold for female-female, though it does for either male-male or male-female. So the downside to filtering out small matches is that you not only eliminate many false positives, you also eliminate genuine matches, though in general these matches would be likely to be less useful, anyway -- though there are exceptions.

As Saetro points out, at the same time that FTDNA may use too low a threshold for X-matches, it does at least allow you to adjust this threshold. However, I'll often work with 5 cM instead of 10 cM. One reason for this is that I've already been able to map my X chromosome fairly well. I know that all of my matches on the X -- with the exception of about 20 cM surrounding the centromere -- came from my maternal grandfather, and therefore from his mother.

As to the 20 cM exception, that was not only from my maternal grandmother, but from her maternal grandmother and not her father or her maternal grandfather. So here I have the segment traced to a specific 2nd great grandmother. At this point, the difficulty of further tracing such matches is compounded by the fact that my 2nd great grandmother's parents were 2nd cousins.

Finally, Saetro's mention of the download capability of FTDNA's chromosome browser is excellent. You can download either the matches currently under observation, or even all matches. Once you've done so, you can sort not only by segment size, but also by segment start or end.

If you create separate sheets in Excel for each chromosome, this allows you to see where every match is located in relation to every other match on the chromosome. Of course, you have two copies of each chromosome (except for the X chromosome, for males), so you can't necessarily always tell if matches that appear to overlap actually do.

However, in my case my father -- who passed away in April of last year -- took the test at 23andMe, and I also uploaded his results to FTDNA. So I can see which of my matches there also match my father, and which do not.

So to recap, I understand your frustration with FTDNA's default threshold for X matches, but you can raise that threshold. And the download function does help make up for the fact that you can't order the X matches by cM length -- though it does require that extra annoying step.