Archiver > GENEALOGY-DNA > 2011-03 > 1299309748

From: Dave Hamm <>
Subject: Re: [DNA] Minimum Parsimony Tree for I1 subclades using STR and SNPs
Date: Sat, 05 Mar 2011 02:22:28 -0500
References: <>
In-Reply-To: <>

Hi David,

Well, I wasn't going to comment on your method. That's because it
is quite complicated to try to get the data set accurately. And tedious,
as well.

If you are happy with "close enough," then I don't have a big problem
with your method. (The "ft2phy" program was my best effort at applying
NIST standards and BLAST searches in order to reflect reality.)

My biggest fear of your method is whether or not you took the time to
error check the data for extra STR's etc. (Such as 464e,f,g and so
on....) A reasonable check on your method would be to compare with Ken's

Other than that, I prefer to use the Mega software in order to get a
pretty tree, that also has a 'balanced' shape.

Regarding the "Outgroup," yes, it would appear that you are using it
correctly. I get the best results by using a haplotype that has the
greatest genetic distance. In those situations I choose a similar
haplotype, but one which falls into a slightly different group. I can
then have an outside group to compare against.

You should find that the "kitsch" program will root the tree for you.
That is probably because kitsch uses genetic distance. Therefore, for a
check on your rooting, you could run your data through the "DNADIST"
program, then the "kitsch" program (for example).

For DNADIST, I would suggest:
select options:

I sequential (if your data is in sequential format)

M for multiple datasets
-D for dataset
- 37 for number (if using 37 marker results in
37 dataset files)
L if you do not want the square distance matrix
D Distance can be Distance (F84, Kimura, Jukes-Cantor, LogDet)
0 (the number zero) If you would like the output in ANSI format

I suppose I should repeat that it is my opinion that you would probably
get the best tree from using Dean McGee's Utility. Plus it's a lot less

I once wrote a blog post about tree comparisons:

With some examples of what that might look like for several software
programs (Kitsch, DNAPARS, DNADIST, TreePuzzle, DNAML, DNAMLK, and Dean
McGee's Utility). I have an example of using DNADIST there. You should
also see that kitsch will root on the haplotype that I use as an
"outgroup" for the other programs.

Yes, I have now been waiting two years for FTDNA to release it's "new"
tree widget.

- Dave

Dave Hamm Franklin, OH

HAM Surname DNA Project Coordinator

On 3/4/2011 6:14 PM, wrote:
> Message: 8
> Date: Fri, 04 Mar 2011 14:21:59 -0600
> From: David Johnston<>
> Subject: Re: [DNA] Minimum Parsimony Tree for I1 subclades using STR
> and SNPs
> To:
> I wish I understood everything I did too ;)
> I got the data off the FTDNA I1 page. Just cut and pasted it all and ran
> some scripts to parse the data.
> There is a category called
> I1 generic (I1-AS (generic))
> I assume that means that it used to be called I1-AS gen but got changed
> to I1 generic. Ken will have to tell you why.
> One thing that I don't understand that well is how to root a tree
> properly. These parsimony codes produce unrooted trees so you have to
> pick a root. Usually I just pick something that I think is outside the
> group and call it OUTGROUP. I think I included a R-haplotype for that
> purpose. Maybe someone can comment on this issue.
> Dave
> On 3/4/11 2:05 PM, Martin Potter wrote:
>> David Johnston wrote :
>>> Have a look. Let me know if anyone has any suggestions for improving
>>> this or comments about the tree.
>> Hi, David,
>> Very interesting and I wish I understood everything that is involved
>> in what you did. (Think I need to go back to school ;-) )
>> I don't pretend to understand Ken Nordtvedt's tree structure, but I
>> noticed that I1-ASgen is missing from your list. Is it called
>> something else in the Tree-MEGA that you presented?
>> ... Martin

This thread: