15 August 1999

Evolutionary Questions Answered By Digital Organisms

by Kate Melville

Evolution is a pain for scientists to study. It takes too long. But scientists may have an answer for speeding up the evolutionary process…

Work by scientists at Michigan State University, Caltech and UCLA has created an artificial world inside a computer, a world in which computer programs take the place of living organisms. They go forth and multiply, they mutate and they adapt by natural selection.

Studying these digital organisms, as reported in the August 12 edition of Nature, offers a chance to test generalizations about how life has evolved.

"These digital bugs are judged on their reproductive performance," said MSU University Distinguished Professor Richard Lenski, an evolutionary biologist and the paper's lead author. "In that sense, they are like organic life. Now the fun begins, because we can start asking questions about them, just like the questions biologists ask about real organisms."

Among the first revelations: Complicated interactions among genetic mutations are very common in these digital organisms, a pattern that also has been reported in real organisms ranging from bacteria to flies. Lenski said this is significant, because it shows that "this artificial world yields some of the same complexities that we biologists see in the real living world, but we have trouble studying these complexities in detail with the real organisms because the genetic experiments get too complicated.

"This is something that has been underappreciated in genetics," Lenski said. "It's been easy to think we understand a mutation when we're just looking at it by itself, but now we can see how it can interact with other mutations, and how its effects change with those interactions."

A less expected discovery: More complex digital organisms are buffered from the damaging effects of multiple mutations than are simpler forms. This finding raises new questions and suggests the need for further experiments with real organisms to see if it also applies to them.

The work is the child of a marriage of biology and computer science.

Lenski, an evolutionary biologist whose work on evolution of bacteria in a test tube has been widely published, has joined forces with Charles Ofria, who recently joined MSU after receiving degrees in computation and neural systems and computer science from California Institute of Technology, and Christoph Adami, a physicist at Caltech who with Ofria designed the computer program to simulate life.

According to Adami, "I think this is the first time we have convinced biologists that artificial life is not just a pipe dream, but is answering some fundamental questions about biology."

The program, called Avida, is basically an artificial petri dish. Ofria and Travis Collier, a UCLA graduate student, created two kinds of digital organisms: simple and complex. The two essentially are cousins: the simple ones' only task is to reproduce. Its more highly evolved relatives, the complex, not only reproduce, but also perform mathematical calculations. Both are rewarded with more computer time.

Avida randomly adds mutations to the programs, thus spurring natural selection and evolution. The team then programs changes in the environment to see how the bugs adapt.

The digital organisms offer an enormous advantage even over the speedy evolution of the rapidly reproducing E. coli bacteria Lenski usually studies. In a Nature publication in 1997, Lenski reported on findings made by looking at 250 different genotypes of E. coli. In a similar experiment run on Avida, the team has been able to scrutinize more than a billion genotypes.

Caltech's Adami says the conclusions are exciting because the "artificial petri dish" approach demonstrates that digital organisms can be used by researchers to answer important biological questions.

"The advantages are that it's very simple and that it abstracts the system as much as possible," Adami says. "It's very difficult to ask very fundamental questions about life with a living system because the living system is very complex after four billion years of evolution."

Understanding the mechanics of how organisms evolve can lead to greater understanding of the inner workings of organisms that can be key to genetic engineering, Lenski said.

"Whether it's bacteria, flies or people, the great challenge of genetics today is dealing with the incredible amounts of data coming from DNA sequencing and new genomic approaches, and then figuring out how all the pieces of the puzzle fit together. Using these digital organisms allows us to simplify things a bit and it speeds up our experiments a lot. They won't answer all of our questions about real organisms, but they may help us shape our ideas and develop new theories."

Moreover, the research also is opening doors into the world of computer science, particularly the science of how computers learn.

"Computer programs are becoming more and more complicated," Lenski said. "We're at the point in some applications where it's hard for the human brain to tell the computer what we want it to do. One area of interest is whether one can employ computer programs that evolve through rewards to do assigned tasks without a human watching each step."