Re: What traits should I look for in a female to breed?

Dear John,

Your question poses some interesting scientific issues. I have to go into a bit of genetics as 
background to answer it.

The genetic material that a person carries is set at conception, and there is no way to alter 
the genetic material that you will potentially pass to your children. A human being's genetic 
material consists of two copies of every gene (except for those on the X chromosome), one 
from that person's mother and one from that person's father. These genes are collected into 
groups on human chromosomes. There are 22 pairs of chromosomes in humans in addition 
to the XX or XY pair.

There are some simple traits that are controlled by single genes. As an example, consider 
the hereditary disease cystic fibrosis. The normal allele (the most common) encodes a 
chloride channel that allows charged ions to pass through cell membranes. There is a fairly 
common set of mutant alleles that do not encode a working chloride channel. A person 
carrying one normal and one mutant allele is a "carrier" with normal health. When this 
person makes gametes, half carry the normal allele and half carry the mutant allele. If they 
mate with another carrier, 1/4 of their children have two normal alleles, 1/2 are carriers, 
and 1/4 have two mutant alleles and are afflicted with cystic fibrosis. In people of European 
descent, about 1 person in 28 is a carrier. These people are healthy and may be slightly 
more resistant to dying from cholera.

There are many complex traits that are controlled by the interactions of two or more genes. 
An extremely complex trait like athletic performance is very difficult to define. We might 
begin by asking the type of athletic performance we are interested in: stamina/work output 
(important for a cyclist), muscle mass (for a weightlifter), reaction time and hand/eye 
coordination (to hit a 90 mph fastball), and so on. If you were to look at the top performers 
for different athletic events you would find a very different set of characteristics. Look at the 
Olympic athletes in the opening ceremonies, for example. Take any one of these people and 
place them in another sport and they would be quite average: the ice dancers wouldn't do 
well at luge and the weightlifters wouldn't play basketball well.

Athletic performance is therefore a poorly defined characteristic. All top athletes are in good 
health, they are well motivated, and they are trained. They don't have obvious genetic 
diseases that affect people in the first few decades of life. Certainly there is variation in body 
type, muscle mass, sprint ability vs. stamina, and so on among top athletes, and they pick 
their sport accordingly.

This is just to point out that what seems to be a simple question is one that is not very well 
formulated from the standpoint of what we know about human genetics.

Despite having sequenced the human genome, we don't have a very good idea of which 
genes contribute to athletic performance. Even if we did, it is very unlikely they they would 
be associated with traits controlling trivial aspects of our appearance (like eye color).

Let's say that eye color is controlled by a single gene (this is almost correct). Suppose 
athletic performance of a particular type is controlled by a single gene (there is no way that 
this is the case). The eye color gene has two alleles: blue (recessive) and brown (dominant). 
The athletic gene has two alleles: good and lousy. If you have two good alleles you are 
great, if you are heterozygous you are average, if you have two lousy alleles you are 
destined to be a couch potato. Because the human genome is a big place, it is most likely 
that these two genes are on separate chromosomes. Even if they are on the same 
chromosome, they may be too far apart to display linkage. If we start with a group of blue-
eyed men who are great athletes (homozygous for blue eyes and good athletic performance) 
and a group of brown-eyed women who are all average athletes (homozygous for brown 
and heterozygous [good/lousy] for athletic performance), we get all brown-eyed children 
half of whom are great and half of whom are average. The next generation has both blue-
eyed (1/4) and brown-eyed (3/4) individuals with all three levels of athletic performance: 
great (9/16), average (6/16) and lousy (1/16). There will be no correlation between eye 
color and athletic performance.

We have started with a contrived example in which all the men are genetically identical at 
these two loci, and all the women are genetically identical at these two loci. This is not 
necessary to predict the outcome in two generations; all we need is the allele frequencies, in 
this example 50% blue and 50% brown for eye color, and 75% good and 25% lousy for 
athletic performance. When the population reaches what is called Hardy-Weinberg 
equilibrium, the allele frequencies predict the frequencies of various genotypes.

Let's say, again, that eye color is controlled by a single gene (this is almost correct). Let's 
say for the sake of argument that there are five genes controlling 80% of the variation in 
athletic performance for cyclists (I'm making that up). The chances are pretty good that 
none of these five genes is on the same chromosome as the eye color gene, or even if one 
of them is, it may be far enough away that it is inherited independently. This means that eye 
color would not be correlated with athletic performance in this more complex and realistic 
example.

You can read more on basic genetics and population genetics in some books online. Go to 
the following site:
 http://
www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=Books

Search terms like "Mendelism" and "population genetics" will lead you to useful chapters in 
"Introduction to Genetic Analysis" by Griffiths et al.

You might also explore this question by looking at the parents of top athletes. The chances 
are pretty good that some of the top athletes have parents that are unexceptional, and that 
some top athletes who have married have children that are lousy athletes. You will find 
exceptions: children of top athletes might be more encouraged to participate in sports, and 
they might find something that they are good at, even if it isn't the same sport. This 
exercise, fairly easy to do because of athlete biographies, will probably convince you that 
athletic ability is not highly "heritable," a term from quantitative genetics which means, in 
this instance, that selecting top performers from a population and mating them does not 
produce offspring that are in the same performance category as the parents.

So to address your questions:

Q1. What traits in a mother produce naturally athletic children?

A1. First, we don't know, and the mother only provides half the genes (except for the 
mitochondria). Genetic variants in mitochondrial DNA are generally associated with terrible 
diseases, not improved performance. Second, there is nothing natural about athletics. 
Performance of a particular athletic task requires general good health and training. It is 
dependent on physical abilities (different for different sports) and also on motivation. 
Motivation may to some extent be dependent on life circumstances and childhood events. 
Finally, it is likely that athletic ability, however defined, is not highly heritable.

Q2. What can I do to manipulate my own composition to ensure superior offspring?

A2. Basically, nothing. The genetic "hand" that you were dealt at conception is what you 
have to play. Since you have two copies of each gene, you might look into your family 
history to see if any of the common genetic diseases are present. If you are affluent, you can 
get yourself tested for being a carrier of cystic fibrosis, the two breast cancer genes (BRCA1 
and BRCA2), familial hypercholesterolemia, and many other genes. If you are a carrier for 
any of these, you can get your spouse tested for the same genes to see if your children 
would be at risk. If that is the case, you might decide to have children by in vitro fertilization 
and preimplantation diagnosis. This won't produce athletes, but at least you can avoid some 
of the known genetic diseases.

Q3. Are there physical traits that I should look for in a woman to mate? 

A3. If you must, you can look for the absence of known disease genes, especially those that 
are the same as yours. If you are a carrier of cystic fibrosis, you might want to check your 
spouse. If she is not a carrier, you have nothing to worry about. If she is, you might want to 
have children by IVF/preimplantation diagnosis. Other physical characteristics (other than 
overall good health) are not predictive of athletic performance.

There are certainly characteristics other than physical characteristics that make people good 
parents. In general, these are learned, not inherited. Good parents might raise healthy 
children that turn out to be concert violinists rather than basketball players. That's not so 
bad. The world needs concert violinists too. There are many things that you can do to give 
children the opportunity to succeed in athletics. Why should it matter if these children are 
yours? If you were to take the thousands of dollars that it would cost to have a couple 
hundred of your genes sequenced and instead donate that to a Little League team, you 
might do quite a bit more good. If you are good at a sport you could volunteer as a coach.

I am only raising this because the human genome project has pointed out how closely 
related all of us are to each other. The average person carries about 25% of the genetic 
variation present in the entire human population. Our parents and teachers tell us that we 
are unique individuals, and that is certainly true, but we are all very close kin, much more so 
than in most other species. The children in your local school are your children too.

You might enjoy reading a good introduction for the scientific layman:
Human Genetic Diversity by Richard Lewontin (1995)

I have used genetic terms in this response; these are defined at:
 http://
www.informatics.jax.org/mgihome/other/glossary.shtml

Yours,

Paul Szauter
Mouse Genome Informatics