I don’t understand what you’re saying. One presumes that paralogous genes would be detected by examining synteny, i.e. similar sequence, different location. This obviously works best if we have a good taxon sample. Visual aid:
A xxxMxxxxxx
B xxxMxxNxx
C xxxxxxxNxx
Supposing that A, B, and C are three species, M and N are similar sequences, and the x bits are flanking sequences that can be aligned among species. M and N are obviously paralogs, if only based on their simultaneous presence in species B or, even if species B were not noticed, their different locations in species A and C.
I don’t know how the Ensembl database worked, but one must assume that an attempt was made to sort orthologs from paralogs, with orthologs being counted as one gene and paralogs as two. Is that not the case?
Don’t understand that. These are relationships, not singular characters. Paralogous to what? Orthologous to what? What is called the same gene in different species is considered orthologous. What is called a different gene in different species is considered non-orthologous, which may be paralogous or non-homologous; the paper doesn’t separate the two. My suspicion would be that most “new” genes in that diagram result from duplication and that most lost genes still have paralogs in the taxa they’ve been lost from. Do you think that’s not true?
I don’t know how to extract the information. I’m just going by the ubiquity of gene duplication in evolution. Consider teleosts alone: they have a whole-genome doubling in their history, and diploidization must have proceeded quite rapidly.
That doesn’t account for the nested hierarchical structure of the data. Sorry, mate.
What waiting time? How is this relevant?
What does sequence space have to do with it?
What magnitude of gene changes? You don’t know how big the changes are, unless you mean the number of gains and losses rather than the magnitude of changes.
Is there not? What do you even mean by “conditions” here?
Ok, excellent, so gene gain and gene loss explains the diagram. Your point of contention with all this is the mechanism by which genes are gained and lost, which you think is undefined to date. Would that be accurate?
Okay I see what you mean. They really can just be paralogs simply because they’re still similar but found in a different location. I had trouble figuring out how they distinguished species-specific genes from orthologues in a way that still made it possible for the species-specific genes to actually be paralogs of shared genes. I don’t know why but I had somehow gotten it into my head that any similar gene shared between two species regardless of location would be counted as an ortholog, which is why I thought the species-specific genes on the Howe diagram couldn’t be diverged duplicates. I get confused.
I see now how both synteny and similarity works together with alignments from multiple species.
Wow. That makes this whole argument with Bill over that diagram all the more stupid and ridiculous. They could literally all just be duplicate genes with the vast majority being tandem duplications even having simply retained their functions. As they say in the legend to figure3 in Howe et al:
Orthologue genes shared between the zebrafish, human, mouse and chicken genomes, using orthology relationships from Ensembl Compara 63. Genes shared across species are considered in terms of copies at the time of the split. For example, a gene that exists in one copy in zebrafish but has been duplicated in the human lineage will be counted as only one shared gene in the overlap.
So the extra gene resulting from duplication in the human lineage will be counted as species-specific because there isn’t a corresponding one in the other species.
I actually missed that one. Are they referring directly to tandem duplications? Because in that case you couldn’t tell which was the ortholog (i.e., the original copy) and which was the paralog. If, however, the duplicate were inserted into a new spot, you would be able to recognize it as a paralog. Now, in the case of the whole-genome duplication in the zebrafish, both copies would be orthologous to the single copy in the ancestral genome but paralogous to each other. Confusing, and I don’t know how Ensembl would treat them.
They aren’t specifically mentioned in the paper no, but I see the problem with tandem duplications. In that case it would make sense to just say that both copies in species A to be orthologoues to the one gene in species B. Checking up on it, from what I can gather that’s also what they do in Ensembl, where they’re apparently called one-to-many orthologoues: https://www.ensembl.org/info/genome/compara/homology_types.html
Okay. Then a species-specific gene that is a duplicate would have to be a duplicate not found in a similar position in any of the three other species. It seems to me this implies of the species-specific genes that are duplicate genes they’d largely owe to the activity of transposable elements then.
I really like that paper - not because I agree with the conclusion, but because of the rigorous methodology for testing the hypothesis. Even for a person who doesn’t like the conclusion, they would do well to use the same methods to test ID hypotheses (there is plenty more data to be tested!).
Hi Matt
Gene gain and loss is a possible explanation. To become an explanation it needs a mechanism to account for the changes.
Separate creation of each animal family (humans, mice, chickens and zebra fish) is also a possibility given the observed differences in the genes of each animal family.
Nope. Doesn’t explain the nested hierarchy in the data. And you can’t call that an explanation, because you don’t have a mechanism for creation of a particular pattern of gene presence and absence.
Common descent alone does not explain the gene pattern in the tree (nested hierarchy) you created. Common descent alone does not explain the Venn diagram. The pattern is not the tie breaker here.
Yes, special creation means the arrangement is the product of design. Now you can retreat to methodological naturalism to eliminate special creation but that only leads us to potentially faulty conclusions.
Yes it does. It doesn’t explain the reasons for gains and losses, but it entirely explains the pattern.
Yes it does. It explains why there are only a tiny number of gene patterns that require more than one change. Nested hierarchy, in other words.
Yes, and that explains nothing. Any arrangement could be the product of design, but that means design can’t explain any pattern we see, in the sense of fitting the data better than some other hypothesis. Nor is special creation a mechanism, since you have no idea what happens. On the other hand, we know how both gene loss and gain can work. Those are actual mechanisms.
Ok, I’ll give it another shot in my own words to see if I’m grasping your position: a combination of gene loss and gene gain explains the Howe diagram only if we have a mechanism or mechanisms to account for the changes. Closer?
The reason for the gene gains and losses is what needs to be explained to move forward from the assertion “the only explanation for the nested hierarchy is common descent”.
If you are limited by methodological naturalism this is true. The idea is that design can explain any pattern says design can explain the nested hierarchy so you cannot eliminate it as a cause unless you invoke methodological naturalism. Since this eliminates another possible cause without a strong explanation for what we are observing eliminating special creation seems premature.
Bill, can you summarize the reasons and references you have already been repeatedly offered for gene gain and loss, and clarify why you do not consider them to be viable explanations?
This is just not true. A nested hierarchy exists regardless of just what caused its elements. You are free to postulate that God personally poofed and zapped various genes at various times, but the poofing and zapping doesn’t explain the pattern. Only common descent explains the pattern, so God must have poofed and zapped within a tree of common descent.
If you are limited by the actual meanings of words this is true.
Not the case. You don’t understand what “explain” means in science. It means that the data fit the explanation significantly better than they fit some alternative. Since design can fit anything perfectly, after the fact, design can’t fit anything better than an alternative, so it can explain nothing. Now, particular hypotheses of design can indeed explain something, but they have to be hypotheses that entail certain expectations of the data and are incompatible with others. Just saying “design” does none of that.
I would claim that special creation does have expectations, and that absence of a nested hierarchy is one of those expectations. Therefore we can eliminate it as an explanation for nested hierarchy.
1.The waiting time to fixation problem per the Behe Lynch discussion.
2.The large sequence space of genes that makes them very likely to degrade over time with random change.
3. How can these large gene changes occur without a high level of purposeful organization and still build a functioning animal?
Animals in the Howe have very different gene patterns. I personally think that the nested hierarchy is a label with a very wide tent and as Winston showed design can generate this type of pattern where conservation of resources is involved.
The Venn is the hard pattern to explain as there are so many differences. Many that do not follow a hierarchal pattern.
