Heredity - The key to paternity determination

People usually resemble their parents. Sometimes more, sometimes less. Some families have typical characteristics so that it is easy to tell that somebody is a relative. The reason for that is heredity. Our DNA is a mix of our parents' genes, and that's why we resemble them more than an unrelated person. However, we can not trust just physical appearance to tell that two persons are parent and son. Physical resemblance is not accurate to establish paternity, but DNA is.

Similar but different

The DNA that is packed inside the nucleus of every human cell is spread among 46 chromosomes and is called the GENOME. Our genome contains about 28,000 genes plus an enormous amount of non-genic DNA with many different functions. All women and men are part of the same animal species, so our genes are essentially the same and almost identical in structure. However, there are regions of the genome that vary among people. Some of these regions vary a little, but others vary a lot, which means that there is a large number of variants in a population.

Got a marker?

Geneticists call 'marker' to any region or part of the DNA that can be identified and analyzed. The more variable is a region of the genome, the better it is for being used as a 'marker' in paternity determination. We call these 'hypervariable markers'.

Establishing paternity
Let's say we pick up a marker for which there are 10 different 'classes' (or variants) in a given human population. Two any persons have one chance in tenth to share the same type of marker. In a population of 100,000 people, each of the ten variants is present in about 10,000 persons.
Now let's try to use this marker to tell whether a child is the son of an alleged father. Father has type 1. If the child has type 1 as well, we have a match! Does this mean the man is the father? Not necessarily. Remember that many other people share the same type but are not related. Now, if the child has one of the remaining 9 types, then we can say that the man is unlikely to be the father.

Making it accurate

The method seems to work but not very conclusively. Can we improve it? Certainly, by including more markers in our analysis. If we look at two markers, each of them having 10 different variants in the population we will have less people sharing the same combination. If we increase the number of markers, the chances of finding two persons sharing the same variants get lower and lower. If we now repeat the test using 10 markers, we can be pretty sure that a child that has the same variant as the man for each markers is actually his son. One can argue that an unrelated person can have the same variants just because. This is theoretically possible, but exceedingly unlikely.



Making it perfect
So if the more the better, could we use hundreds of markers to get more accuracy?
We could, but the study would become too expensive and the difference in accuracy is not worth the cost. Experts have determined that using 16 markers provides extremely high levels of accuracy while keeping the costs reasonably low.