How do we apply evidence of Common Descent?

[T_aquaticus]

@colewd

Specifically state the test and why the results show ancestry. What results would cause rejection of common ancestry?

A while back, @Dan_Eastwood posted this slide:

Across the top are the different markers (e.g CSF1PO, TH01) which are short tandem repeats (STR’s). On the left hand side are the number of repeats in that marker which is defined as an allele. Below each header for each genetic marker is the frequency of the allele in the human population. Using CSF1PO as our example, 5.3% of the population has a CSF1PO allele with 7 repeats, 6.0% have an allele with 8 repeats, and the most common allele is one with 12 repeats found in 29.8% of the population. The slide appears to be cropped, but you can see that there are multiple markers each with multiple alleles.

So let’s do the math. Here is an example of what a DNA profile might look like, keeping in mind that the human genome is diploid so we have two alleles for each gene. I will list the gene and the number of repeats in the allele.

CSF1PO: 9, 12
TH01: 6, 8
TPOX: 7, 10
D5S818: 10, 13
D7S820: 9, 11

Let’s say another DNA sample shared the first allele in each pair (9, 6, 7, 10, 9). What would the chances that two random people would share those 5 alleles? All you have to do is multiple the allele frequencies together. This would be:

(0.037 * 0.124 * 0.017 * 0.070 * 0.109) = 5.9510948e-7 or 1 in 1,680,363.

[note: I am pretty sure I have the math wrong in a fundamental way since it probably requires using nCr or a comparable formula, but it is going to be similar to what I have above.
If someone has the time and knowledge perhaps they could be kind enough to fix my math for me.]

Obviously, the chances of two people sharing lots of markers due to random chance is really, really low. The better explanation for sharing that many markers is common ancestry.

[edit: the last paragraph was a mess]

[colewd (Bill Cole)]

Thanks T

[colewd (Bill Cole)]

@T_aquaticus

Obviously, as you add more markers and those markers are shared between the two people the chances of them sharing that many markers becomes less and less likely that they share those markers through random chance. Instead, common ancestry is the much better explanation.

This appears validates Pauls point that as we go back in time and less DNA similarity is shared common descent becomes harder to validate. I think Joshua’s point to focus on primate common ancestry may be right.

[T_aquaticus]

@colewd

This appears validates Pauls point that as we go back in time and less DNA similarity is shared common descent becomes harder to validate.

This would only be true if evolution and common ancestry were responsible for those similarities and differences. A designer would not be limited by time. A designer could take DNA sequences from wildly divergent organisms and put an exact copy of their genes into a single organism. In fact, humans do this all of the time.

[Timothy_Horton]

@T_aquaticus

This would only be true if evolution and common ancestry were responsible for those similarities and differences. A designer would not be limited by time. A designer could take DNA sequences from wildly divergent organisms and put an exact copy of their genes into a single organism. In fact, humans do this all of the time.

IDers continually ignore this critical point. A designer could mix and match and produce any combination of genetics over life’s history, literally trillions of different combinations. Evolution by common descent can only produce one very distinct pattern. When we observe the fossil and genetic records we observe that one very distinct pattern.

The ID explanation is…to date they don’t have one.

[Dan_Eastwood]

@T_aquaticus

[note: I am pretty sure I have the math wrong in a fundamental way since it probably requires using nCr or a comparable formula, but it is going to be similar to what I have above.
If someone has the time and knowledge perhaps they could be kind enough to fix my math for me.]

You are on the right track, but there’s some information you don’t have yet:

Actually, it’s more than that … more than I can easily describe from memory.

Here are my Powerpoint slides from the brief presentation I did ten years ago:
https://mcw.box.com/s/l1j1grp22x1xo1wqqxt7wpz8h03gfxu0

And a spreadsheet to do the calculations (enter data on first tab, assumptions and results on the second:
https://mcw.box.com/s/c4vfbipxr4xqnodwik6i3qhvdbduyer8

I ask that you not reshare these without my permission.
Those links will expire in a few weeks.

[Mercer (John Mercer)]

@colewd

This appears validates Pauls point that as we go back in time and less DNA similarity is shared common descent becomes harder to validate.

Not at all. It just means that we assay different sequences for different expected separation times.

[colewd (Bill Cole)]

@Mercer

Not at all. It just means that we assay different sequences for different expected separation times.

Not sure what you mean here. Can you expand on this?

[Mung]

@colewd

Not sure what you mean here. Can you expand on this?

The further back in time you go the weaker the phylogenetic signal becomes. so we have to find DNA sequences with a strong signal.

[Dan_Eastwood]

@colewd

This appears validates Pauls point that as we go back in time and less DNA similarity is shared common descent becomes harder to validate.

There is some truth to this. The farther back we look, the more branches that need to be simultaneously estimated, and there will be less power to resolve true structure. I can’t say just how limiting that is without researching it.

I can think of any number of things that would make the model harder to fit, and from the Jeremy Brown I understand that do don’t model certain types of evolution very well yet. There is a famous maxim in statistic:

All models are wrong, but some models are useful. —George Box

Meaning that we rarely know when we have discovered the true model, or if any true model exists, but it can still be good enough to help answer useful questions. The Richard’s/Brown paper/webinar show some of the statistical difficulties involved, and illustrate that it’s difficult to determine exactly where turtles branched off from other reptiles/crocodilians. The models confirm a tree structure for most of the data, and best alternatives where the fit is less certain.

My educated guess is that the data itself must strongly show a pattern of common descent, and if it doesn’t then these models should not work at all.

[Mercer (John Mercer)]

@colewd

Not sure what you mean here. Can you expand on this?

I can, but I won’t until I see some indication of good faith from you.

@Mung

The further back in time you go the weaker the phylogenetic signal becomes. so we have to find DNA sequences with a strong signal.

Not my point. If you’re correct, what exactly is “the phylogenetic signal” and how exactly is it weak in a relevant way for you and a fish?