Hacker News Reader

Comment by surrealize | Hacker News Reader

Ach, this is terrible. Genome size is influenced by a number of factors, including:

- Generation length (faster-reproducing organisms tend to have smaller genomes)

- Transposons (sections of DNA that use cellular machinery to copy themselves to multiple places in the genome; they're usually viruses that got incorporated into the host germline some time in the past)

Genome size isn't anything like a direct measure of "complexity" or "development"; some plants have genome sizes that are orders of magnitude larger than the human genome.

The worst part is that they take genome size sampled in the present day across multiple organisms, and then they assume that some of those present-day genomes represent genomes from particular points in the past. So the x-axis in the regression is basically a guess.

I'm not a statistician, but I imagine that fitting an exponential distribution can make the x-intercept sensitive to data errors.

View on HN

> Genome size isn't anything like a direct measure of "complexity" or "development"; some plants have genome sizes that are orders of magnitude larger than the human genome.

What makes you think those plants aren't more genetically complex and developed than we are? A lot of our complexity is non-genetic.

> I'm not a statistician, but I imagine that fitting an exponential distribution can make the x-intercept sensitive to data errors.

You mean exponential growth, not an exponential distribution, and exponential growth never hits the x-axis.

> What makes you think those plants aren't more genetically complex and developed than we are? A lot of our complexity is non-genetic.

The real problem is that "complexity" and "development" in the context of evolution aren't well-defined. So the way that they place genomes on the x-axis is really arbitrary.

> You mean exponential growth, not an exponential distribution, and exponential growth never hits the x-axis.

Right, exponential growth. And yeah, the horizontal line that they're extrapolating back to isn't the x-axis, it's the y=1 line. The point remains: roughly speaking, it seems like small changes to the curve could produce large changes to the x-coord where y=1, since the curve is so close to horizontal there (on a linear-linear plot), right?

I imagine they're using genome size (number of bases in exons) as their measure of complexity, not something ill-defined.

I don't think it's sensitive the way you say, since they're basically just linearly extrapolating the logarithm, but I haven't read the paper. If you somehow manage to shoot high or low by a factor of 100, that's only a temporal error of six or seven doublings. The bigger issue is that they're extrapolating over orders of magnitude that we have no direct evidence about.

> I imagine they're using genome size (number of bases in exons) as their measure of complexity, not something ill-defined.

They're taking genome size and trying to relate it to time. And the way that they're doing that is to take organisms from the present day, and imagine some of them in the past. The less "developed" or "complex" an organism is, the further in the past they put it. That process is the flawed part.

In other words, the y-axis (genome size) is well-defined, but the mapping of organisms onto the x-axis (time) is not.

Regarding the first point: those plants and lungfish etc with massive genomes aren't more phenotypically more complex. They just contain more duplication. If anything, it's likely a slight impediment to reproduction, but not enough to get selected against.

I'm not sure about this. Plants synthesize a whole lot of biochemicals no animal does, for example, including a lot of amino acids. They can't run away from insect parasites or swat them with their tails, so they have to be able to poison them — without killing the plant or, say, the birds that spread their seeds. There's a lot more phenotypic complexity in corn than you can see without a chemistry lab, probably including things we haven't discovered yet. Do you work in botanical bioinformatics, or are you just guessing?

The problem with your argument is that most plants don't actually have larger genomes. Just some of them. So, if we found a plant with a small genome that synthesized a large range of chemicals, than we could simply assume that the plants with larger genomes aren't getting that kind of benefit. And that is indeed the case. Nobody has found anything phenotypic in these large-genome plants/fish that shows them to be more complex or anything.

My PhD is in Biophysics, my BA in Biochemistry/Molecular Biology, although I don't work in botanical bioinformatics, my education provides me with the tools to make rational decisions based on data.

I see. I guess you're right, then. Thanks!

Reply on news.ycombinator.com