[Taxacom] DNA homologies

[pierre deleporte]

At 14:05 27/09/2006 -0400, John Grehan wrote :

> > A morphological homology statement is necessarily loaded with theory
>
>In a philosophical sense that is true,

great news  ;-)

>Me talking. Molecular biologist refer to random mutations, so given that
>they are random there seems to be no necessary basis for the
>distribution of differences between taxa to track the sequence of speciation.

with permission Sir, this is shear nonsense
any heritable character change, at random or not at random, can label a 
clade in a cladistic analysis [NOT an opinion, just logics]

>The technical issues of algorithmic procedures do not come under my
>expertise so I will leave that to others to argue about.

this is not the crucial point, you can align by hand (see below)
except that a basic "expertise", i.e. a simple introductory course in 
current phylogenetic methods, would of course be necessary

>However, I do point out that there seems to be a contradiction between 
>cladistic
>requirements for primitive states in outgroups and the assumption of
>continuous molecular divergence in all lineages

this classic assumption is only implemented in phenetic analyses of 
molecular (or any other) data, not in cladistic/parsimony analyses of 
molecular (or any other) data

note also that there is strictly no logical contradiction between the 
presence of some "primitive" (plesiomorphic) states in outgroups and 
continuous divergence in all lineages (there are also 
"primitive"-plesiomorphic states in the ingroup, as you know perfectly 
well), and that your adding the term "molecular" in the sentence changes 
nothing to this elementary logical point
[still NOT a matter of opinion]

>that seems to be the underlying principle for viewing molecular similarity 
>as a measure of
>phylogenetic relationship.

again you are describing phenetic analysis, not all current molecular 
analyses (just ask any student passing by in the corridor, for control)

your rhetoric strategy (or hardly conceivable mistake) is now clear enough 
for me
you are persistently playing the game of ignoring that molecular analyses 
are in no way "obligatorily phenetic", neither in the data set nor in the 
phylogeny inference procedure, i.e. ignoring that classic 
cladistic/parsimony molecular analyses are current in the litterature
(if not exactly your pet "compatibility analysis", at least standard 
cladistic analysis - just a mouse click away, i.e. just some character 
deletion away if you prefer hand work)

if you effectively read current litterature, you must know this perfectly 
well - check any recent issue of a phylogenetic journal on your office 
shelves for control

listen, I'm so kind I do it for you, just stretch out my arm...:

not to mention the Journal "Cladistics" (of course you regularly find at 
least some three or four cladistic molecular analyses per issue in there), 
just take Systematic Biology, vol 55 n° 4 august 2006

p. 677-684, you certainly have read the paper by Wortley and Scotland : 
"The effect of combining molecular and morphological data in published 
phylogenetic analyses"
The method is parsimony (= cladistic analysis), and 26 selected papers 
published between 1998 and 2004 are examined - the result is increased 
resolution, not increased support

please listen, John: this is not only cladistic analysis of molecular data, 
this is combined cladistic analysis of molecular and morphological data, 
thus, as explicitly stated in the paper, using only "cladistically 
informative" molecular and morphological data, treated exactly the same way 
of course, because they are treated together at the same time via the same 
cladistic logics from data matrix constitution to optimal tree inference
- strictly no phenetic in sight, quite unfortunately for your general 
argument that cladistic analysis of molecular data do not exist on this 
planet... (see just below for "the spell of algorithms")

I really fear you have to accept such facts, or accept the rejection of 
your counter-factual paper [NOT censorship at all: the effective 
publication of the above mentioned paper, as well as the 26 selected ones 
mentioned in there, is enough proof that they were not censored for 
disobedience to molecules-all-mighty!]

> > The point is that if morphology is problematic in one or more ways, 
> then so too is DNA analysis.

Agreed, no problem, but the specific point at stake is beyond this trivial 
consideration: namely your extensive and persistent misrepresentation of 
phylogenetic analyses;
the fact that some molecularists are occasionally propagandists is no 
argument for using the same counter-propaganda style in the place of 
serious work

symptomatically, your general charge against "algorithms" for molecular 
alignment is surprisingly naive:
algorithms are not evil spirits, they are human-made tools that can 
perfectly implement your very own logic a consistent way, and simply help 
you saving time and avoiding some goofs, as any elementarily trained 
biologist should know at the beginning of the 21st century

but the algorithm spell is very easily exorciseable: please just perform 
alignment yourself by hand, so that you wilk "empirically see" and "observe 
each character claim directly" [to quote your own terms], as any student 
does in its very first basic training, and so that you will not accuse evil 
"algorithms" of distorting your analysis,
and then check by yourself whether or not "the algorithm" gives even 
approximately the same result as yours; not just "in general", but 
specifically for the Human / Pan / Pongo DNA?

so that you can quit the sterile discourse "there seems to be", "I am not 
sure", "I leave it to others" and "I leave it to our respective opinions" 
[logics is precisely NOT a matter of opinions], and you can rationally make 
up your mind directly: it's so simple that you can do it if you really want...

now the Q question is: will your own hand-made, hence strictly 
"algorithm-free" alignment of anthropoids DNA, provide even approximately 
the same result as "algorithm-made" ones? (It's a boring job for complete 
DNA, but just start doing it on "promising" sequences and see what happens)
- if not, it's a real scoop! you're on for an exclusive front page in 
Nature and Buffalo News, and bound for Glory
- but if yes, your whole point simply crumbles down for the specific 
phylogenetic problem at stake

conversely, your projected paper will be of hardly any interest at all if 
you don't do this very simple thing: demonstrate a bias (if any) in 
molecular analyses in Human-Pongo and consorts, which your own nice 
hand-made approach could surmount, instead of reiterating extremely vague 
generalities, the worse for their heavy burden of misconceptions

particularly, your point that cladistic analysis of molecular data is not 
perfomed is simply counterfactual, and as such not worth publishing at all 
- so, what is left?

I suggest it's great time for some quite elementary work, not propaganda

good work and good luck (sincerely)

Pierre



Pierre Deleporte
CNRS UMR 6552 - Station Biologique de Paimpont
F-35380 Paimpont   FRANCE
Téléphone : 02 99 61 81 63
Télécopie : 02 99 61 81 88