More Recent Comments

Thursday, October 21, 2010

Homology, Structural Homology, and Being a Little Bit Pregnant

Misuse of the word "homology" is one of my pet peeves.1 I used to regularly complain about it on talkorigins and I still challenge our graduate students when they talk about "70% homology" or something that's "highly homologous." If you don't understand scientific terminology then it's very likely that you don't understand the concept either.

John Wilkins has pointed me to a couple of good articles2 on the proper use of the word "homology': Distant homology and being a little pregnant and 2010 Homology High-Low Count.

The first article explains why the term "structural homology" should be banned from the scientific literature. The correct term is "structural similarity."

The meaning of homology is on my mind lately because I'm grading essays that critique Jonathan Wells' book Icons of Evolution. Student have to pick one of the chapters and analyze the arguments used by Wells to attack evolution. One of the chapters is "Homology in Vertebrate Limbs" and it's one of the most difficult chapters because Wells highlights the frequent misuse of "homology" in the scientific literature.


1. I have many. It's what happens when you get old.

2. I wonder if he's doing this on purpose—posting a list of provocative articles in the hopes that someone else will do all the work?

11 comments :

-DG said...

A comment I raised in the comments on the Byte Sized Bio article, and which I'll re-iterate here, is the defense of what is usually meant, in my experience, by the term distant homologs. It isn't a qualifier on the word homolog in the same way and isn't intended to represent a degree of homology. Rather distant is a reference to the phylogenetic distance between the homologous sequences.

That's a perfectly legitimate use. It's merely a shorthand for a more tortuous sentence like "homologs of distant phylogenetic relationship" or similar.

Now if people misuse it to mean something along the lines of the dreaded "percent homology" that's something to take issue with, I certainly would. But within molecular evolution papers I see "distant homologs" all of the time, and always in the context of explaining their evolutionary distance, which is often important for proper context of the analysis being performed.

Larry Moran said...

Why not just say that the two proteins/genes/structures are distantly related?

One of the problems with "distant" is that it's very subjective. When I use that adjective, I'm referring to things that last shared a common ancestor more than a billion years ago. In that case, there's often some ambiguity about whether the proteins/genes/structures are actually homologous or not.

Whenever I encounter the term "distantly homologous" in the scientific literature, I automatically interpret it to mean "possibly not homologous" because that's what is usually turns out to be.

-DG said...

Larry: Like I said, in the area of the literature I find myself in, it seems to be used interchangeably with distantly related. Sometimes you'll see one and sometimes the other, also always in cases where the homology is clear and the entire paper is phylogenetically based.

That is obviously a very different case from, for instance, it being used in the structural literature referring to things in the same superfamily, fold class, etc.

DK said...

Because I think that protein folds are monophyletic, I naturally insist that "structural homology" is a perfectly valid phrase that makes all the sense in the world.

A large majority of structural biologists probably feel the same way.

Larry Moran said...

DK says,

Because I think that protein folds are monophyletic, I naturally insist that "structural homology" is a perfectly valid phrase that makes all the sense in the world.

Then you are wrong on two counts. The speculation that all proteins with the same fold share a common ancestor is just that, a speculation. Whether those proteins are homologous is disputed, as you well know. Saying that they are homologous is going far beyond the evidence.

Secondly, the "evidence" that you are relying on is "structural similarity" and not the conclusion of homology. To call it "structural homology" is begging the question.

A large majority of structural biologists probably feel the same way.

Unfortunately, you are correct. We can now add structural biologists to the list of scientists who only think they understand evolution.

John S. Wilkins said...

Regarding footnote 2; I am teaching, and therefore unable to raise sufficient neural power to write anything useful. But the hindbrain still functions well enough to identify interesting things. I just can't do any analysis due to the overload on my frontal cortex.

So the answer is "yes".

DK said...

Secondly, the "evidence" that you are relying on is "structural similarity" and not the conclusion of homology. To call it "structural homology" is begging the question.

First, you know that it's a bit more complicated than just similarity. Second, the conclusion of homology is routinely made based on sequence similarity. So it's the exact same situation that you call "begging the question". All of this is highly tautological, in fact.

Also, I think terms are needed for people to understand each other. When I write "distant homology" you know from the context what exactly I mean. For me, that's good enough. Terms do not exist for people to argue their correct usage incessantly.

We can now add structural biologists to the list of scientists who only think they understand evolution.

Oh, c'mon! It's not like binary concept of homology is a difficult one to understand. Sheesh. Many people simply don't care for your definition! They recognize that "homology" literally means "sameness" and use the term accordingly. Why do you have a problem with that? You know the definition of "homonym", do you? :-)

Larry Moran said...

DK says,

Second, the conclusion of homology is routinely made based on sequence similarity. So it's the exact same situation that you call "begging the question".

Except that we don't call it "sequence homology." That's the difference.

With sequences, homology is based on sequence similarity.

All of this is highly tautological, in fact.

It's not the least bit tautological if you use the terms correctly.

Repeat after me ... "similarity is used to infer homology." When two structures/sequences are sufficiently similar we conclude that they descend from a common ancestor. They are homologous.

Anonymous said...

Guilty!
I once used the terminus "highly homologous" in a paper of mine.

I should have known better and I will not do it again !

The problem with this kind of missuses of technical terms is, that you too frequently read them.
And in a sloppy moment you might use them.

An other problem is that evolution is hardly a topic in the education of biologists anymore (was it ever?)

Thanks for the reminder.
Cheers

DK said...

Repeat after me ... "similarity is used to infer homology." When two structures/sequences are sufficiently similar we conclude that they descend from a common ancestor. They are homologous.

And how do we know that similarity can be used to infer homology? Well, we do - from inferring homology by other means first and finding that it frequently corresponds to similarity. Homology implies similarity and similarity implies homology. Chicken or egg, huh?

Larry Moran said...

DK says,

And how do we know that similarity can be used to infer homology? Well, we do - from inferring homology by other means first and finding that it frequently corresponds to similarity. Homology implies similarity and similarity implies homology. Chicken or egg, huh?

Nope.

There are some clear examples of similarity that aren't due to homology. The examples are called convergence.

The eyes of molluscs and vertebrates are the classic example in the textbooks. I also think there are may proteins with similar folds that are not homologous. The similarities are due to convergence on a stable fold.

Let's say that after careful study you conclude that certain jaw bones in fish are homologous to the bones of the inner ear in mammals. The evidence you collected leads to the conclusion. I don't see a chicken/egg problem there. Do you?