An Easter Test

This is just something to get you thinking over the holiday, to make sure your brains don’t totally ossify. There is a specific reason for asking this: I want to relate it to a blog post I read last night. And, hey, you’ve got nothing better to do than think about phylogenetics whilst you’re digesting your chocolate, have you?


Imagine that you’ve done the following experiment. You have take a single isolate of a phage (a bacterial virus), split it into 5 populations, and grown each of them on one of two bacterial species (two on a strain of E. coli and three on a strain of Salmonella typhimurium), and a higher temperature than they are use to. You culture these populations for 11 days, and find that their fitness in their new environment is higher than their ancestor’s, but fitness on the other host is no better than the ancestor’s (evidence for local adaptation). You also sample and completely sequence one isolate from each population.
So, what would you expect the estimated phylogeny from these sequences to look like? Remember, the true phylogeny is a star. Just to help – yes, there is a signal in the data.
As an added bonus, you cultivate one of the E. coli-grown strains on E. coli, and take one of the S. typhimurium-grown strains and split it into 3 populations: one you continue to grow on S. typhimurium, the other two you grow on E. coli. After 22 days, you sample an isolate from each population. You find that the fitness of the isolates on their new hosts has increased, so that all isolates have roughly the same fitness on the host they have been growing on. Again, you sequence the whole genome of each isolate (hey, these are phages so it’s not too difficult!).
Again, what would you expect the estimated phylogeny from these sequences to look like? And yes, there is a signal in the data.
There are “right” answers, but I’m as interested in seeing whether and how people analyse this and come to their answers. I’ll comment a bit about them when a few people have had their say, and can also answer any questions. There’s one particular reason for this, plus a couple of other aspects that I (at least) find interesting.

About rpg

Scientist, poet, gadfly
This entry was posted in Uncategorized. Bookmark the permalink.

17 Responses to An Easter Test

  1. Henry Gee says:

    I’m not sure if we have enough information. Is it necessary to know the ancestral host of the phage?

  2. Bob O'Hara says:

    No, it’s a single isolate, and the conditions are novel for it.

  3. Henry Gee says:

    Sorry, still confused:
    find that their fitness in their new environment is higher than their ancestor's, but fitness on the other host is no better than the ancestor's
    If the new environment is E. coli and Salmonella, what’s the ‘other’ host? Or does ‘new’ mean ‘coli’ and ‘other’ mean Salmonella?

  4. Bob O'Hara says:

    Sorry. The other host is E. coli for the populations grown on S. typhimurium and S. typhimurium for the populations grown on E. coli.

  5. Mike Fowler says:

    How do you measure fitness? Is it something related to a phage’s ability to run marathons or something about survival and reproduction?

  6. Bob O'Hara says:

    It’s growth rate, I think it’s a standard microbiology measure. I assume they do it in a medium with excess resources, so it’s asymptotic growth. Which is close enough to fitness in the original experiment (I don’t want to go into the gory details: it’s a bit off-track).

  7. Henry Gee says:

    Am still confused by the first part. Is there a step in the first experiment where the phage, cultured on one bacterium, is then swapped with a new culture of the other bacterium?

  8. Bob O'Hara says:

    Not in the first step. It’s just taken from a stock, I actually don’t know what it was taken from, it’s described as a “wild type”. The fitness of the original is quite low in the new conditions on both hosts.

  9. Cath Ennis says:

    42

  10. Richard Wintle says:

    The phylogeny looks like a poorly-rooted tree, with really small and statistically non-significant branch lengths.
    Do I win? Did I even understand the question?
    Henry, can you help me with the buses now?
    /coat

  11. Eva Amsen says:

    Does it look like this? (The dots are just there for spacing)
    \|/
    .|
    /.\
    This is probably totally wrong, but I’m confused by the original thing being a star shape, I think. Otherwise I would have just drawn a line or something Y-shaped.

  12. Henry Gee says:

    I remain completely confused by this. OK, let’s say the ‘stock’ is growing on something that is neither E. coli nor S. typhibusstation – we have, therefore, an outgroup, against which the new stocks form a monophyletic group. But then
    You culture these populations for 11 days, and find that their fitness in their new environment is higher than their ancestor's, but fitness on the other host is no better than the ancestor's
    Which means that they are fitter than their ancestor, but then again, they aren’t.
    I think I must have missed something.

  13. Eva Amsen says:

    I’m confused by there being five things but only two different conditions.

  14. Bob O'Hara says:

    Sorry for not replying earlier – I’ve been kidnapped by some developmental biologists, to talk about Web 2.0 and listened to talks about things like GoPro49.

    Which means that they are fitter than their ancestor, but then again, they aren’t.

    It’s host specialisation: it evolves to grow better in the conditions its growing in. But that doesn’t affect it’s fitness in a different environment (i.e. the other host).

    I’m confused by there being five things but only two different conditions.

    There were 5 replicate populations (3 in one condition, two in the other).

    Does it look like this? (The dots are just there for spacing)

    Yes. Now can you (or someone, if anyone else is still reading this) explain why?

  15. Richard Wintle says:

    I shall revise my earlier statement:
    The phylogeny looks like a poorly-rooted starfish, with really small and statistically non-significant branch lengths.

  16. Eva Amsen says:

    YAY! Uhm the three branches that are in one group are for the three that grew on one strain, and the two on the other side are for the two that grew on the other strain. Within one subgroup they’re going to be less different than between these two groups, because those are in totally different environments (different selective pressure) so the distance between the groups got bigger. And because fitness on the other host is the same as on the original, it really was just the host that they grew on that determined the difference.

  17. Bob O'Hara says:

    Eva – that’s what happened. Richard would be right if there hadn’t been strong selection, but the conditions the populations were put into meant that the sequences were attracted to regions with high fitness, and that signal was picked up in the phylogeny.
    I’ll try and find the time to write a full explanation of this this week.