DNA file format change affects GEDmatch results

[JOlson]

The GEDmatch algorithm only compares SNPs that exist in both kits being compared. You are correct - It DOES affect accuracy. When only 40% of the SNPs line up, it will be less accurate than when 100% of the SNPs are present. Of course, using 1cM thresholds is also less accurate than using 10cM thresholds....

[JOlson]

Wet finger estimate indicates about 95% complete given kits from both parents. If only one parent is available, then the phased kits are probably about 80% complete....

[JOlson]

Need more information. What is the degree of the match? How many segments and cM of each segment match on autosomal and X?

[JOlson]

GEDmatch has had a 3D Chromosome browser for over a year. It displays relationships between matching segments for up to 10 kits on all 23 chromosomes. The 3D view can be rotated for closer examination. The accompanying matrices give additional details about the segment matches....

[JOlson]

Yes, I'm talking about matching segments. Consider this simplified example:

Person A:
A A A A
C C C C

Person B:
C C C C
C C C C

Person C:
A A A A
A A A A

Person A matches B, and Person A matches C,...

[JOlson]

Triangulation

Just because A matches B and A matches C, doesn't mean that B also matches C. This may help explain:

Triangulation

.

[JOlson]

I don't think you would have all those Single Segment matches if they were identical twins. Identical twins would appear to be the same individual in a DNA comparison. If they are loaded on GEDmatch, you will easily be able to see the difference by enabling the graphic on the one-to-one comparison....

[JOlson]

There was a server failure. I have been able to get a few utilities back online so new batch results could be made available. Not sure how long a permanent fix will take. At least a couple of days, maybe longer depending on how serious the problem turns out to be.

John...

[JOlson]

FTDNA reports the "+" strand. NCBI doesn't always do that. I think they show either "+" or "-".

Just remember:

A <=> T
C <=> G

so in your case, CC is the same as GG....

[JOlson]

Don't use X results when calculating relationships.

Also, don't use 7200 cM. Use something closer to half that amount....

[JOlson]

I've been able to do some more analysis on this problem. It seems to affect a relatively small percentage of uploads, but goes back several months. A disproportionate number of the affected kits seem to be conversions (mostly Ancestry.com), but I can't really say for sure.

The effect...

[JOlson]

The format of FTDNA files recently changed. No-Calls were previously represented as "--", but are now shown as "00". The GEDmatch upload software was not properly interpreting the new code for no-calls, and was instead inserting a hard break which caused a mis-match at every location...

[JOlson]

marietta,

Also, have you tried its replacement?

[JOlson]

One time, I was talking to a guy who worked in one of the test labs. He said somebody submitted a sample of what he thought was probably a dog.

Most other animals don't even have the same number of chromosomes as humans.

http://en.wikipedia.org/wiki/List_of_organisms_...

[JOlson]

19% Ancestry.Com
28% FTDNA
4% 23andMe V2
47% 23andMe V3
2% 23andMe V4

John...