Overview of Early Avida Work

Posted November 24, 2015 by Joshua Nahum in Information, Review / 1 Comment

Back in 2005, I was tasked to pick an article written by a science journalist and write a summary of it for my AP Biology class. During those days, I would frequent the library to read science magazines like Scientific American and Popular Science. Little did I know, the article I found would profoundly shape my career to this day. I chose the cover article for the February 2005 edition of Discover titled: “Testing Darwin” by Carl Zimmer. The article described the work done at Michigan State University to study evolution in a completely different system than DNA-based life. Carl Zimmer, who continues to be my favorite science journalist, described experiments studying the rise of complex features, evolution that generated diverse ecologies, altruism, the benefits of sexual recombination, among other ideas, but all using the digital evolution platform, Avida. From that point on, I always had a version of Avida installed on my laptop, where I’d play with the settings and see what happened.

Since then, there’s been a great many important and interesting scientific works done in Avida. But, I wanted to highlight here some of the early papers that fueled my fascination back then in 2005. These aren’t every paper on Avida from that period, but the ones that grabbed my attention.

  1. The Avida user’s manual (1998) – This is the first description of how to install and run Avida, it is very much out-dated by this point, but I still have a little heart throb when I think of the genesis file and the text console viewer.
  2. Genome complexity, robustness and genetic interactions in digital organisms (1999) – This paper was the first to make it clear to me the connections between the evolution of artificial and natural organisms. I’d read papers examining epistasis and mutation effects in bacteria, but it was exciting to see that the same sorts of experiments can be done in digital systems. They also found a surprising result: more complex organisms appear to be more robust to mutations.
  3. Evolution of biological complexity (2000) – This paper has some of the first examples of types of analyses that we do in Avida experiments today, like looking at the entropy of specific sites in the genome. By looking at entropy (roughly, the amount of variation present in the population) at a position in the genome, you can learn what parts of a digital organism are heavily conserved versus more variable. They also looked at hitchhiking (where non-beneficial mutations can be fixed because they co-occurred with beneficial ones) and Muller’s ratchet (mutations accumulate if selection isn’t strong enough to remove deleterious mutations before they fix).
  4. Twice as natural (2001) – This is a short (1 page) essay on the need to study evolution with digital organisms.
  5. Evolution of digital organisms at high mutation rates leads to survival of the flattest (2001) – This paper is another biology connection. They found that at high mutation rates, digital organisms didn’t evolve to the highest fitness genotype, but instead to the highest “region”. At high mutation rates, one’s descendants will inevitably have mutations making them different from the parent. If the parent is in a genetic region where it has high fitness, but all nearby mutants are low fitness, most of its children will be badly off. But if a moderately fit parent is in a region where mutants of it are also moderately fit, it will ultimately be selected over the more fragile first parent. This effect is termed survival of the flattest. This effect was subsequently found to occur in natural organisms (most notably in viral quasi-species), leading to a nice pattern of theory -> found in artificial organisms -> found in natural organisms.
  6. Evolution of Stable Ecosystems in Populations of Digital Organisms (2002) – In this paper, digital organisms experienced environments where they could increase their growth rate if they consumed resources. In experiments where resources were limited, organisms specialized and formed complex and stable ecologies that persisted indefinitely, even when mutations were turned off.
  7. Design of Evolvable Computer Languages (2002) – To be honest, this is probably my favorite paper of the bunch. Here at the Devolab, we primarily work with the same instruction set (digital amino acids, if you will) for all of our projects, with only small changes when called for. This paper explores the effects of different instruction sets that affected features like nops (no-operation instructions), different conditional instructions, different numbers of instructions, and so on. It’s easy to just go with the default, but with digital organisms, everything is a knob you can adjust if you want to.
  8. The Effect of Natural Selection on Phylogeny Reconstruction Algorithms (2002) – Phylogeny reconstruction algorithms are used to infer genetic relationships and history given only the genetic sequences of organisms alive today. Because we don’t know the genetics of ancient and extinct organisms, we need to infer from what data we have. This paper applies these algorithms to populations that we do know the history of (because they watched it happen and controlled it in Avida) to examine how well they work in a non-trivial evolutionary model. I often use this as a justification for Avida as it can be used to test biological assumptions and difficult to assess models.
  9. The evolutionary origin of complex features (2003) – This is the best paper to read as your first introduction to Avida. It describes how complex features can and do arise in evolutionary systems. This shows how “irreducibly complex features” (features that can’t function if multiple parts aren’t already present) can evolve despite creationist claims of its impossibility. It also contains a nice, short description of what Avida is, making it a good paper to read as a group/class.
  10. Avida: A Software Platform for Research in Computational Evolutionary Biology (2009) – I would be remiss if I included the 1998 manual without pointing out that a more recent version exists. This is a revised manual of the intricacies of how Avida works and how to use it all in one manuscript. It is mostly up to date, but the source of information we recommend for the most recent version of Avida follows next.
  11. GitHub Wiki For devosoft/avida (present) – GitHub is where we host and develop Avida, and the wiki associated with the repository contains the current information on the workings and usage of Avida.

So there’s my list of interesting papers from 2003 and earlier. If there are other papers you enjoyed that you think should be added to the list, please let me know in the comments. Thanks!


Joshua Nahum

I am a postdoc in the Ofria Lab and BEACON, with an interest in the interplay between environments and the organisms that evolve there in. I'm also an enthusiast in computer science and software development, despite my PhD in Biology.

More Posts - Website

One response to “Overview of Early Avida Work

Leave a Reply