Distance Calculation May Be Misleading

[K. Campbell]

Most of us rely on FTDNA's distance calculation as a measure of closeness. However, use of this number alone may be misleading.

My project has just come across our first example where a pair of results whose distance of 4 show a more recent Time to Most Recent Common Ancestor than a pair of results whose distance is 3. The difference is due to the marker-specific mutation rates used in the FTDNATiP calculations

(NB: all information below is based upon 37 marker comparisons)

For example,

Kit A (44575) and Kit B (22678) are a distance of 4 apart. The FTDNATip calculation of TMRCA shows:
29% with 4 generations
69% with 8 generations
89% with 12 generations
97% with 16 generations

Both kits A and B are a distance of 3 away from kit C (34905). The FTDNATip calculation of TMCRA for both individuals shows:
10% with 4 generations
42% with 8 generations
72% with 12 generations

This result is counter-intuitive, and I believe, based upon the marker specific-mutation rates used in the FTDNATip calculation.

First, I'd request that FTDNA publish these rates so that everyone could understand how the FTDNATip calculation is performed.

Second, I would caution Admins not to make decisions related to distance only.

Third, I'd request that the Genetic Distance report on the Members Page be Changed to having matches sorted by TMRCA. While this report is accurate for what it is, Admins really need something that better reflects the "degree of closeness" if distance isn't the best metric.

Thoughts?

[DMac]

I am not the Admin of our study at FTDNA, but did create and host the web pages where the results are displayed, so I'm not familiar with the tools you may have available. I'd like to see your first suggestion come about. Having the actual mutation rates that FTDNA uses would be most helpful doing comparitive analysis. The second suggestion I would expand to caution against taking any one measure or tool as the 'end all' of your analysis. I use several tools...Dean McGee's on-line comparison utility, Fluxus-engineering's free phylogenetic network analysis software and just basic Excel spreadsheets.

Just like with Medicine, there's a lot of science behind DNA testing but there's a large piece that is more 'art' than science. Getting 'friendly' with the results is as important as any one tool that may be used.

The third suggestion I just don't have an opinion on since I don't use it.

I do have a question about your results. Are the differences between A and B the markers that FTDNA has indicated as faster mutating and maybe the difference(s) between A/B and C are on a slower mutating marker?

[Eki]

Here's a comparison between differen t markers and their mutation rates:

http://www.worldfamilies.net/marker.htm

The way I understood it is that despite some markers mutate more often than others, each mutation takes just one generation to happen. So, for example, DYS388 (mutation rate 0.0005) might have mutated when it was passed from my father to me or it might have mutated 2000 years ago or any time between them.

[DMac]

Here's a comparison between differen t markers and their mutation rates:

http://www.worldfamilies.net/marker.htm

The way I understood it is that despite some markers mutate more often than others, each mutation takes just one generation to happen. So, for example, DYS388 (mutation rate 0.0005) might have mutated when it was passed from my father to me or it might have mutated 2000 years ago or any time between them.

Eki,
I've seen those rates that Doug deduced for FTDNA. My understanding is that FTDNA has apparently changed some/all the rates when they came out with their FTDNATiP in the last year, year and a half. Having the current rates being used would be ideal.

You are correct in your understanding that any mutation could have taken place at any time.

[K. Campbell]

FTDNATip Calculation

In response to a private question about this issue, Bennett has responded:

We treat both 1 and 2 step mutations in the FTDNA tip calculator as single evolutionary event, so if you just count the GD points and then use the TiP calculator you will sometimes not get the same GD. Dr. Walsh suggest that if you are closely related then whether you have a single step of difference or 2 steps of difference it happened as a single movement, ergo he treats it for the purpose of the TiP calculator as a single evolutionary event.

Also since some markers move much faster or slower then others you will see different results based upon which marker has moved.