C Morley's Y-DNA phylogeny Geno 2.0-ISOGG integration

[seferhabahir]

5) I tried several known L21 private SNPs and your tool did not find these. The only one I can remember is L583 (very small scope). This SNP is a son of DF13 via testing but was reported only as L21 (did not show either DF13 or L583). It would be hard to keep up with all these private SNPs (L583 formerly was the son of L21 before DF13 was discovered). Since my interest is these private L21 SNPs for genealogical purposes, I would be glad to assist you with catching a few of these. Many of these are documented at my L21 Private web site:

L583 does not show up in this tool because I have not yet transferred my Geno 2.0 results to FTDNA (there was some concern early on about faulty transfers). It's not necessarily a fault of the tool, since I believe it will only report things based on publicly available Geno 2.0 SNP results or things in the ISOGG tree (neither of which apply for L583). One of these days I will get around to doing the transfer. L583 is now known to be under Z251, and will no doubt be under some others once my Full Genomes results are known.

[RobertCasey]

L583 does not show up in this tool because I have not yet transferred my Geno 2.0 results to FTDNA (there was some concern early on about faulty transfers). It's not necessarily a fault of the tool, since I believe it will only report things based on publicly available Geno 2.0 SNP results or things in the ISOGG tree (neither of which apply for L583). One of these days I will get around to doing the transfer. L583 is now known to be under Z251, and will no doubt be under some others once my Full Genomes results are known.

I just tried one of the four L583+ submissions that show up in the L21 project's YSNP report. The tool returned L21 as the terminal YSNP. I believe it was 201644 which is not tested as DF13+. There are several L583+ and L583- submissions that have tested positive for DF13. This means his tool is not automated for discovery of most private SNPs via analysis of the tested submissions. I am attempting to automate that part of SNP hierarchy.

[Rathna]

Morley 2.0.
How is it possble that Kasdin (N9149) is a subclade of Mendez/Alvarez (150747) and not Vass (40065) when Kasdin and Vass have the same haplotype and different from that of Mendez?
!1R1b1c PF6279/V88 [3/3], PF6281 [3/3], PF6287 [3/3], PF6289 [3/3], PF6290 [3/3], PF6292 [3/3], PF6293 [3/3], PF6295 [3/3], PF6305
[3/3], PF6307 [3/3], PF6310 [3/3], PF6324 [3/3], PF6328 R [3/3], PF6329 [3/3], PF6330 [3/3], PF6332 [3/3], PF6333 [3/3],
PF6338 [3/3], PF6339 [3/3], PF6340 [3/3], PF6343 [3/3], PF6344 [3/3], PF6376 [3/3]
1314 !!!!!!!!!!!!!!!!!140065
1315 !!!!!!!!!!!!!!!!!? 1R1b1c1 M18
1316 !!!!!!!!!!!!!!!!!? 1R1b1c2 V35
1317 !!!!!!!!!!!!!!!!!? 1R1b1c3 V69
1318 !!!!!!!!!!!!!!!!!!1R1b1c–1 PF6304 [2/2], PF6327 [2/2]
1319 !!!!!!!!!!!!!!!!!!1150747
1320 !!!!!!!!!!!!!!!!!!!1R1b1c–1–1 F1389 [1/1]
1321 !!!!!!!!!!!!!!!!!!!1N9149
1322 !!!!!!!!!!!!!!!!!1R1b1–14 L388 R [801/801], L389 [801/801]
1323 !!!!!!!!!!!!!!!!!1N16605
36 Kit N9416 has 1 positive SNP from the R1a1a1b1a3a1a–1 level

[Rathna]

Morley 2.0.
How is it possble that Kasdin (N9149) is a subclade of Mendez/Alvarez (150747) and not Vass (40065) when Kasdin and Vass have the same haplotype and different from that of Mendez?
!1R1b1c PF6279/V88 [3/3], PF6281 [3/3], PF6287 [3/3], PF6289 [3/3], PF6290 [3/3], PF6292 [3/3], PF6293 [3/3], PF6295 [3/3], PF6305
[3/3], PF6307 [3/3], PF6310 [3/3], PF6324 [3/3], PF6328 R [3/3], PF6329 [3/3], PF6330 [3/3], PF6332 [3/3], PF6333 [3/3],
PF6338 [3/3], PF6339 [3/3], PF6340 [3/3], PF6343 [3/3], PF6344 [3/3], PF6376 [3/3]
1314 !!!!!!!!!!!!!!!!!140065
1315 !!!!!!!!!!!!!!!!!? 1R1b1c1 M18
1316 !!!!!!!!!!!!!!!!!? 1R1b1c2 V35
1317 !!!!!!!!!!!!!!!!!? 1R1b1c3 V69
1318 !!!!!!!!!!!!!!!!!!1R1b1c–1 PF6304 [2/2], PF6327 [2/2]
1319 !!!!!!!!!!!!!!!!!!1150747
1320 !!!!!!!!!!!!!!!!!!!1R1b1c–1–1 F1389 [1/1]
1321 !!!!!!!!!!!!!!!!!!!1N9149
1322 !!!!!!!!!!!!!!!!!1R1b1–14 L388 R [801/801], L389 [801/801]
1323 !!!!!!!!!!!!!!!!!1N16605
36 Kit N9416 has 1 positive SNP from the R1a1a1b1a3a1a–1 level

Only because Sam Vass has had a no call in PF6304 (certain because documented by him) and probably also in the other two SNPs (PF6327 and F1389). And so it is demonstrated that the Jewish R-V88+ cluster derives from these Spaniards like Mendez/Alvarez, who belongs to the same haplotype of Arellano, I put on Ysearch from SMGF at the time of Rootsweb (2007), when I said to Vass that his haplotype was an introgression from Spaniards and, even though he is an Ashkenazi Jew, his Y came from Iberia.

P.S. Of course we should be sure that Mendez/Alvarez hadn't a no call in F1389.

[Rathna]

Only because Sam Vass has had a no call in PF6304 (certain because documented by him) and probably also in the other two SNPs (PF6327 and F1389). And so it is demonstrated that the Jewish R-V88+ cluster derives from these Spaniards like Mendez/Alvarez, who belongs to the same haplotype of Arellano, I put on Ysearch from SMGF at the time of Rootsweb (2007), when I said to Vass that his haplotype was an introgression from Spaniards and, even though he is an Ashkenazi Jew, his Y came from Iberia.

P.S. Of course we should be sure that Mendez/Alvarez hadn't a no call in F1389.

Probably the explication is another. From the Itai Perez spreadsheet

https://www.dropbox.com/s/zv0khjlk6g...mpare_view.zip

we can see that Vass has two no calls in PF6304 and PF6327, but has no mutation in
F1389. Then these three R-V88+ agree in all the mutations and only Kasdin has a private mutation in F1389.
It remains these huge presence of PF SNPs, in favour of an Italian origin of the haplogroup and an introgression of R-V88+ from Iberia in the Jewish pool.

[cmorley]

I've released a new version (8 August 2013) of my phylogeny. It addresses some of the comments from this thread.

From http://ytree.morleydna.com/:

8 August 2013
A new version of the experimental phylogeny was released today. It features over 1850 Geno kits. It is the first version of this report to fill in gaps using select non-Geno data. Some of these non-Geno SNPs will be private. Others will turn out to be further upstream, only for the moment appearing to be private because they haven't been widely tested yet.

The new version also features corrections to some of the discrepancies noted in the 26 July version.

I have made some changes to the data structure underlying my experimental phylogeny. Until I make the corresponding changes to my predictor, its experimental predictions will continue to be based on the 26 July 2013 phylogeny.

As a demonstration of this software's versatility, the same algorithm was used to process the Geno data was used to generate a phylogeny from the two-sample Full Genomes demo data.

By request, this and future versions of my phylogeny will be accompanied by a CSV summary file giving kit counts and SNP counts for each node.

Finally, at the end of last month I produced an automated report showing discrepancies and gaps in the ISOGG tree and SNP database.

Rathna raised a good point about kit 40065. The owner of this kit has confirmed that he got no-calls for PF6304 and PF6327. The new version of my phylogeny reflects this. This is a useful cautionary tale: my software will point out potential flaws in the phylogeny, but it cannot always detect no-calls at or just above the terminal SNP. I agree with Robert that FTDNA needs to make changes to its Y-SNP reports. From my perspective, the greatest benefit would come from the reporting of no-calls and the separation of Geno and non-Geno results.

The 26 July version of my phylogeny (and subsequent versions) make deliberate provisions for the SNPs from David Reynolds' list. This enhancement had little impact on the resultant phylogeny; my algorithm was already making adjustments for these SNPs without explicitly being told to do so.

Regarding (3) and (4) from Robert's 24 July post: yes, my predictor accommodates Geno 2.0 and Full Genomes demo data, but the 17 July and 26 July phylogenies only used Geno 2.0 data. The 8 August 2013 version of my report includes a separate phylogeny built from the two Full Genomes demo sets. I am not going to merge these two phylogenies just yet -- the reliability of the Full Genomes calls needs to be better understood, and the numerous discrepancies marked with a superscript F (see, for example, CT in the Geno 2.0 phylogeny) need to be resolved.

This means his tool is not automated for discovery of most private SNPs via analysis of the tested submissions.

That is correct. I cannot assume that everybody wants their submitted data made public. Nor can I assume that all user-submitted data is reliable.

The 8 August version of my phylogeny makes use of select public non-Geno data. Consequently, my phylogeny now includes L583 and many other narrow-scope or seldom-tested SNPs. What do you think?

[razyn]

The 8 August version of my phylogeny makes use of select public non-Geno data. Consequently, my phylogeny now includes L583 and many other narrow-scope or seldom-tested SNPs. What do you think?

I like the fact that it picked up our L484 under CTS4065, no "official" tree has been doing so. 217955 is my cousin, and if you locate him in the DF27 project there are four other L484+ tested people with 111 marker results and the same STR off-modals (a dozen of them). The rest of us haven't done Geno 2 tests, but we have a shared MRCA, so I believe the higher level SNPs (CTS4065, Z295, Z210) are not in question for the group.

Your algorithm also correctly picks up the division after Z295, whereby CTS12074 and CTS4065 (at least) have mutually exclusive descent lines below Z295.

[cmorley]

A new release is available. As before, see http://ytree.morleydna.com/experimental-phylogeny . My preference is that people link to this address, rather than linking directly to one of the PDF files. (This measure prevents people from accidentally using outdated versions.) Of course, if you do this, make sure to mention which version you are talking about.

The new version of this phylogeny incorporates several thousand non-Geno kits.

This new release required additional under-the-hood data structure modifications. Consequently, the classifier is still using the 26 July 2013 version of my phylogeny.

[TigerMW]

Paul,

The August version of my experimental phylogeny was the first to include non-Geno SNPs. For non-Geno SNPs, the algorithm took an "assume it defines a new subclade until proven to be equivalent" approach.

The September version of my phylogeny (see http://ytree.morleydna.com/experimental-phylogeny) now has Z2973 at the same level as M222. The algorithm behind this version of the report is more prone to assume equivalent until proven otherwise. Borderline cases are reported: when a handful of kits are positive for an SNP but negative for an SNP previously presumed to be phyloequivalent, the unexpected negative results are recorded beside the kits in question.

Chris

Chris, thanks for your work. I realize its volunteer work so any advice I have can be discarded with no offense to me.

My suggestion on terminology would be to assume "equivalence" until proven non-equivalent. However, rather than use the word "equivalent" use the term "approximate". I can see why Thomas Krahn used the "approximate" terminology on his draft trees. It does not convey an assumptive "equivalent" nature but it also does not convey an assumptive older or younger positional nature. Approximate is really about as accurate a word as can be used.

[cmorley]

Chris, thanks for your work. I realize its volunteer work so any advice I have can be discarded with no offense to me.

My suggestion on terminology would be to assume "equivalence" until proven non-equivalent. However, rather than use the word "equivalent" use the term "approximate". I can see why Thomas Krahn used the "approximate" terminology on his draft trees. It does not convey an assumptive "equivalent" nature but it also does not convey an assumptive older or younger positional nature. Approximate is really about as accurate a word as can be used.

That's totally fair. Does "approximately phyloequivalent" cover all the bases?