Report from Alife XII: life's origin, and its evolution

When I say 'artificial life', what do you think of? I think of life-like systems in computers, mainly, but at the Alife 12 conference in Odense, Denmark that I am currently at, a large part of the presentation are really about chemistry. Many people might be surprised if they knew just how many people are working on the problem of getting chemicals to behave like life. That is, work on the origin of life is booming. Take a look at the program. A few talk titles:

• Autocatalyses
• Spontaneous Assembly of Cell-Like Structures from Likely Prebiotic Materials: Problems and Prospects
• Light induced Replication and Selection in Peptide Networks
• The Origin of Life by Serial Dilution of a Primordial Soup
• Evolution, Selection and the Metabolic Foundations of the RNA World
• Dynamical Stability of Autocatalytic Sets
• Synthetic (Constructive) Biology: From Vesicles Self-Reproduction to Semi-Synthetic Minimal Cells
• Constructing an artificial self-replication system of genetic information from RNA and proteins

The point here is just that, yes, we don't know how life originated, but people are hard at work on the problem. And they have reason to believe that it can be done in the lab, eventually. As such, the fact that it is very difficult does not mean that the problem is unsolvable. And even less does the fact that some people cannot imagine how it could occur without the word of God (also, how exactly is it that God would have done it?) have any bearing on how it happened. Of course.

In two days I am talking about some work I've done on the NK model: Critical properties of complex fitness landscapes (in EvolDyn V). In anticipation of someone asking what the relevance of this model is, I am thinking about what to say about that in the introduction (which might take up half of the alloted 15 minutes).

My particular implementation of the NK model has 20 loci. That's a lot less than natural systems have, obviously, but on the other hand much more than the usual two that can be modeled analytically. Twenty loci is enough to result in very complicated fitness landscapes, when the loci are set to interact epistatically (i.e., the effect of a mutation depends on the particular genotype of the organism), but also not more than we can completely enumerate, such that the fitness of every possible genotype is known (for N=20 loci there's a little over a million possible genotypes, and we have looked at organisms with N=32, which gives almost 4.3 billion).

Basically, the only way to learn about the global properties of fitness landscapes is to simulate them in a computer. In natural organisms it is extremely cumbersome to survey even a limited area of genotype space. This has, for example, been done in yeast (Costanzo et al., 2010). They did knock-out experiments, meaning that they disabled single genes in pairs, such that the fitness of the organism could be measured for the unaffected wild-type organism, a yeast organism with one gene knocked out, another with a second gene knocked out, and finally a fourth organism with both of those two genes disabled.

It turns out that the bigger the effect of the single mutant is, the more epistasis there is between the two genes that were knocked out. In other words, the extra effect on fitness of the interaction of the two genes is on average larger when the effect of the first gene is large. That tells you something non-trivial about the shape of the local fitness landscape centered on the wild-type genotype. But the really intriguing observation is that this shape around a single genotype is eerily similar to that in the NK model, and the point that I want to emphasize is that this observation validates the NK model to some extent, as a model that does have something to say about real fitness landscapes.

Reference:
Costanzo, M., Baryshnikova, A., Bellay, J., Kim, Y., Spear, E., Sevier, C., Ding, H., Koh, J., Toufighi, K., Mostafavi, S., Prinz, J., St. Onge, R., VanderSluis, B., Makhnevych, T., Vizeacoumar, F., Alizadeh, S., Bahr, S., Brost, R., Chen, Y., Cokol, M., Deshpande, R., Li, Z., Lin, Z., Liang, W., Marback, M., Paw, J., San Luis, B., Shuteriqi, E., Tong, A., van Dyk, N., Wallace, I., Whitney, J., Weirauch, M., Zhong, G., Zhu, H., Houry, W., Brudno, M., Ragibizadeh, S., Papp, B., Pal, C., Roth, F., Giaever, G., Nislow, C., Troyanskaya, O., Bussey, H., Bader, G., Gingras, A., Morris, Q., Kim, P., Kaiser, C., Myers, C., Andrews, B., & Boone, C. (2010). The Genetic Landscape of a Cell Science, 327 (5964), 425-431 DOI: 10.1126/science.1180823