[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Subject Index][Author Index]

Re: Rconstructing DNA (was Re: Dino-fuzz found in amber?)



On Mon, Sep 19, 2011 at 6:23 AM, David Marjanovic
<david.marjanovic@gmx.at> wrote:
> First, are you sure you haven't misunderstood the technique?

Yeah, pretty sure.

> Are you sure it doesn't involve phylogenetic bracketing right from the start, 
> so that you
> need a bird _and_ a crocodile -- or, rather, the ancestral neornithean and
> the ancestral (crown-)crocodylian sequence -- _before_ you start comparing
> the *Tyrannosaurus* aa sequence to anything?

"If we had the crocodile or alligator or caiman sequence the comparison
could be improved."
http://dml.cmnh.org/2011Sep/msg00204.html

"It would make difference, since it could help in philogenetic bracketing."
http://dml.cmnh.org/2011Sep/msg00210.html

"And, naturally, outgroups."
http://dml.cmnh.org/2011Sep/msg00212.html

But we can compare T-rex peptide sequences with the chicken's one -
and proceed to DNA analysis. It will only limit that analysis
somewhat. What we will have is a unrooted tree with two ends - yeah, a
simple line: the ancestor is somewhere in between - assuming a regular
molecular clock, it will be more 'close' to T-rex end. But we could
simplify in an preliminar analysis, and put the ancestor midway in the
line.

Using parsimony we tick the line with the nucleotides changes needed
to go from one end two the other.

As more terminal is added it surely could (and is expected to) change
- the distribution of 'ticks'.

> Because that way, you would _inter_polate. What you did was to
> _extra_polate.

Interpolation is more secure, indeed. But extrapolation could be done.
Nevertheless, what was done is not exactly a extrapolation: we are not
absolute clueless in regard to T-rex DNA sequence, some of information
could be get from the aa sequence; and it is improved with the chicken
sequence.

> Second, it has happened before that molecular biologists used unsound
> techniques. The literature on molecular dating is full of papers that use
> completely erroneous calibration points (and too few of them), and even
> fuller of papers that use all calibration points as maximum ages and thus
> arrive at much too old dates.

Not only molecular biologists, every field is subject to errors (there
are many flawed paleontological papers, for example). And once the
errors are discovered the techniques can be improved.

I'm not saying that because a technique is currently used by experts
it is 100% error-proof. What I'm saying is that the fact that a
technique is currently used by experts, it would be regarded as good
enough: the one that is unsatisfied with the technique is who have the
burden of proof to show that the technique is not good.

>>  The rationale is basically the same as the morphological
>>  phylogenies. We can infer that the most recent common ancestor of
>>  protostomes and deuterostomes had bilateral simmetry. There are many
>>  characteristics that are inferred for the Urbilateria.
>>
>>  Even for Cenancestor or LUCA.
>
> Absolutely not. All conclusions about them are drawn from _inter_polation:
> you compare descendants and draw conclusions about their common ancestor.
> You do not go beyond that.

Sure. But no one is going *beyond*. Just making a back and for
movement. From the descendants, we reconstruct the ancestral state.
Knowing the ancestral state (and part of the descendant state and some
rules about character state), we could infer some other descendant
states.

If in the homologous position, the descendants have the same aa, then,
most probably, it is a shared characteristic inherited from the common
ancestor. Lets say that it is arginine.

We know that arginine is coded (in most of nuclear genes) by one of
the four codons: CGT, CGC, CGA, CGG, AGA, AGG.

In this case, we are looking just for two descendants - [surely it is
better to added more terminals between then (looking at the unrooted
tree)] - if in one end it is used the codon CGT, it is more probable,
then, that in the other end the codon is CGN, since otherwise, it will
requires two steps to go from end to the other: either going from CGT
to AGA/G or going from AGA/G to CGT.

We do similar things when we find a incomplete fossil. We know that
the ancestral dinosaur is a tetrapod with all four limbs (and it is
known by comparative analysis of the descendants). We find a dinosaur
skeleton with three limbs: we could infer that the individual used to
have four limbs (unless we find information that lead us to conclude
other way).

[]s,

Roberto Takata