How theory of evolution is useful: predicting food allergies!

We’ve had some discussions in the past about whether the theory of evolution provides any benefit to society. Here’s one more benefit to add to the list: you can use common ancestry/phylogenetics to predict whether you’ll be allergic to a food you’ve never eaten!

Here’s the approach biologist Christina Agapakis takes:

  1. Sequence the proteins that you know you are allergic to.
  2. Sequence related proteins in foods you are interested in trying.
  3. Build a phylogeny (nested hierarchy). This is the common ancestry step.
  4. Use the phylogeny to predict which foods you will be allergic to, and which you may eat.

Yes, I remember that discussion quite well. The TOE is useful for trivial things such as raising food crops and combatting disease.

That’s not the theory of evolution speaking. It’s merely observed similarities between organisms. Nature doesn’t come with the imaginary lines you call nested hierarchies.

Think about it: from that diagram, if you are allergic to tomatoes, all you need to know is that potatoes are more similar to tomatoes than mint or figs. Which one came first and when it’s irrelevant.

Hi NonlinOrg -

You are making a specific statement that can be demonstrated by research and evidence.

Your statement is that predictions based on similarity measures only, without nested hierarchies, are just as reliable as predictions based on nested hierarchies.

This statement could easily be supported by data. Thus I was rather surprised that you did not supply any.

Christina Apagakis provided her data sources and research methods. You could easily obtain the same data, perform a similarity measure, and compare it with a nested hierarchy measure. What’s stopping you?

Best regards,
Chris Falter


Do they compare against the control of using similarity alone?

1 Like

No, that’s not what I said. All our knowledge including the information used in the study you quote comes from observed similarities. There is no other information.

I just wanted to know why my mom suddenly developed peanut and soy allergies last year :confused:

@NonlinOrg has a point. As far as is explained in this study, they did not do the control of comparing phylogeny to similarity alone. So it cannot be interpreted as evidence for evolution. It is possible the original studies to which she referred did, but I did not have time to find and read them. However, as written here in Scientific American the study is specious

So it is possible in @NonlinOrg’s hypothesis is correct…

The right response to this is to actually do the controlled study. To be clear, however, that sort of work that compares similarity to phylogeny HAS been done elsewhere, with very clear results…

We find that this hypothesis is false…

If you doubt it, you need to provide evidence. But when you look at the evidence, you will find that nature does come with nested hierarchies.

And moreover, you will find that function follows phylogeny closer than it follows similarity. The deficiencies of this study do not negate this observation.


And what is phylogeny based on if not on similarity?

Per Wikipedia: Usual methods of phylogenetic inference involve computational approaches implementing the optimality criteria and methods of parsimony, maximum likelihood (ML), and MCMC-based Bayesian inference. All these depend upon an implicit or explicit mathematical model describing the evolution of characters observed.

Evolution is not an explanation but an assumption that gets to a phylogenetic model. But the model is an output of two inputs (observed similarity and assumed evolution) and hence cannot be used to prove either of the two inputs. Regarding this topic, similarity is the only thing needed to predict food allergies. Evolution or the phylogeny model do not add anything to to this prediction.

1 Like

@NonlinOrg, I think this article may suffice to answer your concerns.

29+ Evidences for Macroevolution: Part 1

Phylogeny is different than similarity. There are poor/incorrect methods of inferring phylogeny that use similarity. But the best methods use nested clades instead, and this is usually what we mean when we talk about phylogeny. It turns out that these two methods actually produces different trees.

So this turns out to be 100% false.

First, these studies do not assume evolution. Rather, like all modeling efforts, they make quantitative predictions with the evolutionary model. We then compare this to other approaches like similarity. The results of this comparison is what tells us which model works better. There is no assumption that evolution is correct here. Rather, we use the data to determine if the theory of evolution helps us understand evolution. It is a direct test of your theory that phylogeny is equivalent to similarity. We find your hypothesis is FALSE.

Second, it is 100% false that the input to phylogeny is just similarity and the assumption of evolution. Phylogeny algorithms require the full sequences, because the need to determine how nested clades cluster sequences. Nested clades cannot be inferred from similarity. I understand that this is all new to you, but your understanding of this is incorrect. You are wrong.

Remember, no assumption is made. The data tells us which model works better. Because similarity and nested clades produce different trees, we can then ask which type of tree gives a better description of our world. Here are three articles that carefully do this comparison, and all come to the same conclusion. Phylogeny (by nested clades) is about two times better at explaining the function of proteins than similarity.

Do not fear the evidence @NonlinOrg. God is the author of all truth, including this.


Thanks, the article you referenced is definitely coherent and well articulated, but not convincing.

In the end they have two arguments (1):
“So, how well do phylogenetic trees from morphological studies match the trees made from independent molecular studies?”
See, morphology is not independent from molecular characteristics.
They counter with “many very different DNA sequences or biochemical structures can result in the same functions and the same morphologies” and “Heredity correlates sequences, even in the absence of functional necessity: There is one, and only one, observed mechanism which causes two different organisms to have ubiquitous proteins with similar sequences (aside from the extreme improbability of pure chance, of course). That mechanism is heredity.”
Just because the model doesn’t fall apart, doesn’t make it anything more than a model. And “one, and only one, observed mechanism”??? Funny considering that we have not observed said mechanism and no one is interested in even trying to test it: Saying “this is the only way” is equivalent to sticking one’s head in the sand.

…and (2): “Although it is trivial to classify anything subjectively in a hierarchical manner, only certain things can be classified objectively in a consistent, unique nested hierarchy. The difference drawn here between “subjective” and “objective” is crucial and requires some elaboration, and it is best illustrated by example. Different models of cars certainly could be classified hierarchically—perhaps one could classify cars first by color, then within each color by number of wheels, then within each wheel number by manufacturer, etc. However, another individual may classify the same cars first by manufacturer, then by size, then by year, then by color, etc. The particular classification scheme chosen for the cars is subjective.”
Who classifies cars/beings by color, year and size? But if you do the proper classification, you will put the 2000 VW Beetle after the 1970 model, not before it. And in the end you end up with an objective grouping that matches the development history. Yet cars are all designed implements.

OK, I’ll buy that. Apagakis derived benefit from an analysis based on evolutionary thinking and methodologies. However, her goal was not to show the superiority of evolution-based methods to similarity-based methods.

Exactly. When phylogenetic models consistently yield more accurate predictions than similarity-based models, it is strong evidence that phylogeny is more accurate description of the ground truth.

If I am not mistaken, NonlinOrg, parsimony refers to the notion that the cladistic structure that minimizes the number of branches is preferred. With a large number of taxa this could be very difficult to compute by brute force, which is why Markov Chain Monte Carlo Bayesian inference is used to derive the most parsimonious cladistic structure. Distance/similarity measurements are vastly easier to compute, so MCMC is not used for them.

Long story short, these approaches are calculating nested hierarchies, not distance/similarity measurements.

1 Like

Look at this @NonlinOrg ! @Chris_Falter is admitting his original point was wrong. The allergy study did not actually test evolution vs. other theories, and he recognizes this now. That is the sign of humble thinker, and make me trust him all the more.

I’ll look forward to you doing the same sometime.

1 Like

Google: “A clade is a grouping that includes a common ancestor and all the descendants (living and extinct) of that ancestor.” So Clades presuppose evolution.
I define Similarity more broadly to include all info available from morphology to DNA. I think you mean to narrowly include only morphology.

If so, what are the alternative hypotheses tested? Please do not include random distribution here.

I wrote something else. Please review.

Rest assured I do not fear the evidence. I already said many times that True, False or Revised, either way Evolution doesn’t affect my religious views.

I am not trying to win points here. Just to learn something new and to test my hypotheses. The fact that I continue the discussion should be taken as an appreciation of your thoughts.

1 Like

This is not what I am referring to. I am referring, instead, to a specific pattern in sequences. In this case, it has nothing to do with morphology. Of course, morphology falls in nested clades too, but that is a different conversation. We are talking about sequences here.

Just read the papers I linked to and the original post. It is very clear.

As I have explained several times, the alternative hypothesis tested is that similarity predicts function better than phylogeny. This exactly your hypothesis.


@NonlinOrg you would understand more by reading the whole thread, but this is a key post.


I agree that classifying by color is a poor choice. However, it may be educational to look deeper at this analogy.

The main difference between a developmental ‘tree’ of cars and the tree of life is that we see characteristics get invented and, within a few years, spread to all cars, regardless of their line of ‘descent:’ anti-lock brakes, electric keys, etc. Multicellular creatures don’t operate this way; they only, if you will, descend according to kind or lineage. (Bacteria play by different rules for gene transfer.)

Human design messes up the tree. It is still possible to do a car tree, but you have to take into account and ignore ‘common design’ elements like the spread of new technology or new regulations.

The tree of life does not have these kinds of elements. The closest you could come would be convergent evolution cases, but it is quite obvious when we look closely at these instances that the underlying mechanisms and processes are entirely different even if the phenotypic results are similar. Instead, the pattern of descent, nested hierarchy, is quite stark and mathematically clear.

I’ve said before, if people want to accept that the pattern looks like descent while actually being just the way God created, I have no further argument with them. My problem is when people deny reality to say that the pattern of the tree of life was invented and no better describes biology than a bunch of sticks in the ground.