Posted on 01/23/2004 7:18:12 AM PST by PatrickHenry
STANFORD, Calif. – Early humans migrating from Africa carried small genetic differences like so much flotsam in an ocean current. Today’s studies give only a snapshot of where that genetic baggage came to rest without revealing the tides that brought it there. Now researchers at the Stanford University School of Medicine have devised a model for pinpointing where mutations first appeared, providing a new way to trace the migratory path of our earliest ancestors.
The study was led by Luca Cavalli-Sforza, PhD, emeritus professor of genetics, who has spent most of his career tracking the evolution of modern humans. Much of his current work involves following mutations in the Y chromosome, which is passed exclusively from father to son, as humans migrated from Africa and spread to the rest of the world during the past 50,000 years.
These mutations, most of which cause no physical change, tend to appear at a constant rate, providing a genetic timer. For example, if a population has 10 mutations after 50,000 years of evolution from the common ancestor in Africa, then the fifth mutation probably arose 25,000 years ago. But where was the population located at that time? Until now genetics hasn’t had an answer.
“If we know the time when a mutation arose we know something. If we also knew the place we’d know almost everything,” Cavalli-Sforza said.
With the help of senior application software developer Christopher Edmonds and statistician Anita Lillie, both researchers at Stanford, Cavalli-Sforza built a computer model to simulate how mutations spread in a migrating population. The results of this work are published in this week’s online issue of Proceedings of the National Academies of Science.
The group reduced the world’s continents to a simple rectangular grid. They populated the first few squares with computerized human populations and gave those electronic villages realistic rates for population growth, migration and mutations. The inhabitants had more than one child, on average, and those offspring could migrate to any neighboring square as long as it wasn’t filled to capacity. This population growth filled the initial squares to capacity and pushed the computerized people to migrate at a constant rate across their rectangular territory until the next space was filled.
When a mutation appeared within a population, descendants reproduced and migrated at the same rate as other individuals. Most of the mutations, however, simply disappeared due to chance.
Those mutations that stayed in the population until the simulation ended showed one of two patterns. If the mutation appeared in a heavily populated area, it had a lower chance of surviving for many generations or reaching high numbers. In these cases, the mutation remained extremely rare in the local population.
If a mutation appeared in a person at the edge of the migration front where the population was scarce, the mutation was more likely to spread through the population. The mutation-carrying person multiplied and the offspring migrated, taking the mutation to neighboring squares. If these neighboring squares were previously unoccupied, the mutated person had a high probability of reproducing and passing along the mutation. The mutation itself remained most common in the migratory wave front, a situation Cavalli-Sforza refers to as “surfing” the migratory wave.
Over the course of 64,000 simulations, the group came up with a model for finding a mutation’s origin. First they identified the mutation’s farthest edge – corresponding with a boundary such as the ocean or mountain range in human populations. Then they calculated the average area of where the mutation is distributed – called the mutation’s centroid. According to the models, the centroid is about half the distance between where the mutation arose and where it ended up.
In at least some simulations, the mutation no longer existed in the population where it first arose. Without the group’s way of estimating distance, there might be no trace of the mutation’s place of origin. Now they can generate a dated “we were here” sign to place on the route of human migration.
I don't think so, unless you tamper with the results. The reason I asked "if you can program a model giving the same result as the one in the article, with the same mutation and migration rates, but with a starting point of only 6,000 years in the past" is because I don't think you can.
As I said, I'm not a programmer. But I've done modeling with a spreadsheet. If I know the starting quantity (say $100K, the amount invested), and the ending result desired ($1 million), I can diddle with the variables (rate of return and time) until I get the desired result. But I can't do that if the rate of return and the time are already known -- not unless I delve into the bowels of the underlying software and produce a false result.
In the case of the article, the factors are known (starting point, ending result, and the rates of mutation and migration). The input was known. The output was known. What was missing was the model itself. The challenge was to create a model which fit the data. That's a very specific challenge, and no matter how good a programmer you may be, it doesn't mean, as I think you're claiming, that you can conjure up just any old model and make it fit the data. You may be a great programmer, but I don't think you understand what the article is all about.
Abstract:
The ability to infer the time and place of origin of a mutation can be very useful when reconstructing the evolutionary histories of populations and species. We use forward computer simulations of population growth, migration, and mutation in an analysis of an expanding population with a wave front that advances at a constant slow rate. A pronounced founder effect can be observed among mutations arising in this wave front where extreme population bottlenecks arise and are followed by major population growth. A fraction of mutations travel with the wave front and generate mutant populations that are on average much larger than those that remain stationary. Analysis of the diffusion of these mutants makes it possible to reconstruct migratory trajectories during population expansions, thus helping us better understand observed patterns in the evolution of species such as modern humans. Examination of some historical data supports our model.The full text of the paper can be read accessed from that link, but will require a PNAS subscription or a one-time $10 fee.
I think you'd really enjoy the book, "Guns, Germs, and Steel", by Jared Diamond. The author does a good job of piecing together the rise of human civilization and (early) technology, which started around 15,000 years ago. He makes a good case for why agriculture arose in some human societies but not others, why some peoples developed cities and armies before others, and so on.
It's a fascinating and illuminating book, and I guess I'm not the only one who thought so, since it won the Pulitzer prize.
I am. And I'd like to say that you're grossly misrepresenting the nature of computer modelling.
Yes, you get an analysis of the problem, which is *not* the "any result you want" you implied.
If I model a calculation of the square root process, and I use it to get a result showing that the square root of three is 1.7320508, is that only because I *want* it to be that, or because it *is* that? Can I "make" the square root of three to be "any result I want"? No, I can not.
Computer models are used to analyze processes and tell the people using them what happens under certain circumstances, not so that *they* can tell" the model what happens.
If that's how *you* model things, you've been doing it all wrong.
I repeat, you're grossly misrepresenting the nature and use of computer modeling.
I've been a programmer for thirty-two years. How about you?
I still haven't communicated my point. It's not that they got the desired output. That was already known. This isn't a case of some idiologically-motivated programmer saying: "Look, my model predicts global warming. Head for the hills!" I agree with you, anyone can do that. Even I could do that. What's remarkable in this case is that they fashioned a model. The model is the "theory," built around all the known data points, which attempts to show how the output was achieved.
If, using the same data, you can produce a different model, then you have a competing theory. The evidence of future digs will determine which is correct. Either humans migrated and mutated as your model predicts or they didn't. Reality will have the final word.
I think this is great news. It means we can ALL now apply now for reparations!
Let's hope it stays on topic. A Crevo thread would actually be more civilized than an amnesty thread these days.
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.