A Match On Relative Finder But Not On Family Finder

[shandy4473]

Good evening everyone,

Has anyone received a match on 23andME's Relative Finder, but not on Family Finder, assuming both parties tested on Family Finder. In addition, has anyone gotten a paper trail confirmation of their Relative Finder match?

I am asking because I matched to someone on Relative Finder but not on Family Finder even though both of us tested on Family Finder. We both share 18cM and 3441 SNP on our chromosome 2. Gedmatch saw us a match as well on the same chromosome.

Thanks
Steve

[Brunetmj]

I am on both places but see no duplicate people as far as I can tell. Since this site uses first and last names and 23andme doesn't it is hard to tell for certain.
I found a confirmed third cousin once removed here.

[nolnacsj]

I have two matches who match me on both RF and FF. However, as the previous poster said, without knowing the names of the majority of my 23andMe matches, I cannot compare all my matches.

Judy

[Ann Turner]

I analyzed one very quirky case where a 12 cM segment was found with 23andMe (based on 500,000 SNPs) and FTDNA's Affy chip, but it disappeared with FTDNA's Illumina chip. It involved a microdeletion in one of the parties, which split the segment into two smaller segments, each below the threshold for declaring a match. The Illumina chip had more markers in the vicinity and showed that a single mismatch with the Affy chip was not just one of those occasional genotyping errors.

However, 18 cM should still show up, even if it's split into two smaller segments.
FTDNA has some proprietary rules about SNP blocks -- perhaps your segment fails one of those unknown rules. I'd be glad to analyze this in more detail if you and your match are willing to share your raw data from 23andMe and FTDNA with me. My e-mail is [email protected].

[shandy4473]

Thanks Ann

Thanks Ann. I will ask my match to send her raw files from FTDNA and 23andME. Once my match sends them - I will send my raw files as well.

Thanks
Steve

[EdwardRHill]

I just did the test at 23andme and have matched some of my FF matches there. I did have three new matches father, son, daughter that we confirmed Common Ancestors with paper trail.

[shandy4473]

Email from Helpdesk.

I analyzed one very quirky case where a 12 cM segment was found with 23andMe (based on 500,000 SNPs) and FTDNA's Affy chip, but it disappeared with FTDNA's Illumina chip. It involved a microdeletion in one of the parties, which split the segment into two smaller segments, each below the threshold for declaring a match. The Illumina chip had more markers in the vicinity and showed that a single mismatch with the Affy chip was not just one of those occasional genotyping errors.

However, 18 cM should still show up, even if it's split into two smaller segments.
FTDNA has some proprietary rules about SNP blocks -- perhaps your segment fails one of those unknown rules. I'd be glad to analyze this in more detail if you and your match are willing to share your raw data from 23andMe and FTDNA with me. My e-mail is [email protected].

Hi Ann. Happy Holidays. I finally received a response from the FTDNA Helpdesk = "Dear Stephen,

Thank you for your email. I reran the comparison and I can confirm that the kits are not matching a long enough block or a great enough total to be considered closed by the Family Finder algorithm. It is possible that the other matching algorithms allow for greater variance and thus resulting in a confirmed match, where we are not seeing it.

Darren Marin" - Now I am starting question Relative Finder's accuracy.

I now have a new case where FTDNA views a new match with my dad but not with his mom. However GedMatch claims there a match with BOTH my father and his mother. I have emailed the FTDNA helpdesk for another manual comparison.

Thanks
Steve

[djparlette]

My shortest cM matches here are 7.73 cM. At 23 and Me they are 7cM to be on Relative Finder. So yes, I do have people who have tested at both places but do not show up here, but they are below 7.73cM's. ( and we do have confirmed paper trails.) I also have about ten people who show up as matches to me on both sites.
Donna

[shandy4473]

Response from GreenSpan

I analyzed one very quirky case where a 12 cM segment was found with 23andMe (based on 500,000 SNPs) and FTDNA's Affy chip, but it disappeared with FTDNA's Illumina chip. It involved a microdeletion in one of the parties, which split the segment into two smaller segments, each below the threshold for declaring a match. The Illumina chip had more markers in the vicinity and showed that a single mismatch with the Affy chip was not just one of those occasional genotyping errors.

However, 18 cM should still show up, even if it's split into two smaller segments.
FTDNA has some proprietary rules about SNP blocks -- perhaps your segment fails one of those unknown rules. I'd be glad to analyze this in more detail if you and your match are willing to share your raw data from 23andMe and FTDNA with me. My e-mail is [email protected].

Here is what Dr Greenspan says in response to my case

Hi Steve

Sometimes this is more 'art' then science.

Big is a relative term. Here's what we find when we compare the two kit you asked about:

MatchSerialNo: 194506-3425
TotalCentimorgans 13.590817
LongestCentimorgans 11.424491
SecondLongestCentimorgans 2.166325
SharedSegments 2


Yes you do have a longest block of 11cM, but the next block (which we call a supporting block) is very very small, and is more of what we call a IDS block rather then a real IBD block. that tends to reduce the value of the long block itself, since if it were genealogical it would be reasonable to expect that you'd have supporting blocks which these two samples don't show.

The minimum amount of matching DAN that we allow is about 20CcM, and the two samples you had me run are quite a bit lower then that.


Best Regards

Bennett Greenspan
President
Family Tree DNA

https://www.familytreedna.com

"History Unearthed Daily"

[shandy4473]

2nd and 3rd response from Dr Greenspan

Here are two more responses from Dr GreenSpan relating to my case. It looks like what truly matters is what's in the matching algorithm!!!

2nd response - "Hi Steve

Looks to us that this is probably an Identity by State block. If related it's prior to 3rd great Grandparent level. From a genealogy standpoint Id say that they are not related, they just have some DNA common."

3rd response - "Hi Steve

In the screen shot I showed you the relationship ID was 17, meaning non related. With about 20 cM in common we begin to show relatedness in a genealogical time frame."

Best Regards

Bennett Greenspan
President
Family Tree DNA

https://www.familytreedna.com

"History Unearthed Daily"

[JamesBianco]

I have 3 people who tested with 23andMe and FTDNA and who match me in both. All three are French Canadian matches, and we share at least 5 ancestrial couples (10 Ancestors) at the 8th to 9th cousin level, and even more if you go further back. Trouble is that there was so much inbreeding in French Canada, it is difficult to determine which of these ancestors gave us our common DNA.

Both my parents and both my children tested, the matches hold up with them as well, although some are on my dad's side and some my mom's.

So it is a legitimate match.

[Ann Turner]

The minimum amount of matching DAN that we allow is about 20CcM, and the two samples you had me run are quite a bit lower then that.

I had worked through the raw data for Steve and his match, and there were no discrepancies at all in the segment (over 4,000 SNPs, so that's quite impressive). The GEDmatch example was for Steve's grandmother, and did show a couple of discrepancies, but that's well within the allowance for genotyping error.

The message from Bennett sheds new light on the reason for the lack of a match at FTDNA. I'm wondering if the "rule" is appropriate for all types of ancestral backgrounds.

[shandy4473]

Email Mr. Greenspan

I had worked through the raw data for Steve and his match, and there were no discrepancies at all in the segment (over 4,000 SNPs, so that's quite impressive). The GEDmatch example was for Steve's grandmother, and did show a couple of discrepancies, but that's well within the allowance for genotyping error.

The message from Bennett sheds new light on the reason for the lack of a match at FTDNA. I'm wondering if the "rule" is appropriate for all types of ancestral backgrounds.

Hey Ann - You may want to email Dr Greenspan directly. He seems to be a very open and responsive person.

[Ann Turner]

Hey Ann - You may want to email Dr Greenspan directly. He seems to be a very open and responsive person.

Yes, I have -- Bruce Walsh also.

[thetick]

Interesting I have Family Finder matches with:

Shared Largest Predicted
22.64 -- 8.64 -- 5th Cousin - Remote Cousin
20.93 -- 16.48 -- 5th Cousin - Remote Cousin

I would be very curious what the curt offs are. Based on Greenspan's comments my second match above should have not matched?