Interpreting matches with low cM values and high SNPs (or vice versa)?

4

On GEDmatch's autosomal one-to-one report, a match contains a column for cM and a column for SNPs values, for each of the matching chromosomes. Normally you look for high cM values as an indicator of genetic distance (and to some degree, generational distance). But I also understand that it's important to take into account the number of SNPs in relation to the cM value (the greater the SNPs, the better it is).

But, in my research, I've discovered high cM values with very low SNPs (for example, a common value that I find is: 14cM / 128SNP). On the other hand, I have very low cM values with very high SNPs (3.2cM / 4968SNPs).

I've not found these contrasting amounts useful for establishing valid matches so far (and often delete them from my database). But I'm wondering if I should take these "extreme" segment matches more seriously -- do they have any value? How should I interpret them?

TJK

Posted 2018-11-28T19:37:47.867

Reputation: 133

Answers

2

Ignore the SNP counts

Centimorgans are a weighted unit of measurement. The cM number depends entirely on the location of the segment and has little to do with actual length.

The SNP number is the actual length (more or less since all commercial DNA tests with matching only test a fraction of total SNPs).

Sure, the SNP number can be useful to advanced researchers, but it's not anything you need to worry about. Focus on the cMs. And ignore any under 5cM (preferably anything under 7cM), unless you're doing a very specific task where small segments can give you information (intermediate level stuff).

Cyn says make Monica whole

Posted 2018-11-28T19:37:47.867

Reputation: 2 426

Your answer could be improved by adding links to posts explaining why small segments are usually ignored, such as Blaine Bettinger's A Small Segment Round-Up: https://thegeneticgenealogist.com/2017/12/29/a-small-segment-round-up/

– Jan Murphy – 2018-11-28T22:39:49.510

However, at GEDmatch it is important to note small SNP counts. Different companies test different SNPs, and if there is a low SNP count for a match, say under 500, then it is a low quality match with a good chance of being a false positive no matter how many cM the match might be. – lkessler – 2018-11-29T04:51:16.830

In one of Dianne Southard's webinar's, she suggested that 15 cM is actually a better threshold - that segments below 15 cM are about 80% likely to be IBC and above 15 cM about 80% likely to be IBD (I'm sure it's actually more subtle that that). This also seems to be in line with my personal experience. – cleaverkin – 2018-12-04T19:57:08.437

I'd be interested to see more on that. She is going against all the genetic genealogy wisdom and leaders that I've seen. 7 is pretty standard with 5 common. I may have seen something higher than 7, but not higher than 10. 15 is rather shocking, actually. – Cyn says make Monica whole – 2018-12-04T20:03:20.967

Are you sure that's not the threshold for the largest segment? Most of us in endogamous communities say 20 for the largest, but 15 is not unreasonable, especially for non-endogamous. The 5 or 7 cM segment minimum I'm talking about is for any segment used in adding up the total cM match. For a close match, there would need to be one or more segments significantly larger than that. – Cyn says make Monica whole – 2018-12-04T20:03:25.077

Maybe I misremembered, it might be largest segment. It was one of the Legacy Family Tree webinars, I'll see if I can dig up a reference. In my case, I was referring to matches with 20-25 total cM, so largest would also fit that scenario. – cleaverkin – 2018-12-04T20:07:42.777

OK, I found the reference and I did mis-represent it. It was from Diahan Southard's "Genetic Genealogy: Advanced" webinar at Legacy Family Tree Webinars. The figures she claims are that at 7 cM, segments are about 80% likely to be IBC, at 15 cM about 80% likely to be IBD, and at 10 cM about 50% either way. She was referring to all segments, not just largest segments. – cleaverkin – 2018-12-06T07:41:41.163

Some of us do not buy the IBC vs IBD methodology (the naming of it and the assertions about segment size) but that's a longer discussion for another post. Thanks for coming back with the update. – Cyn says make Monica whole – 2018-12-06T15:28:45.140