How DNA Paternity Testing Works

We take DNA samples by wiping cells off the inside of the cheeks of each patient using swabs. No blood. It is painless. (These are called "buccal cells" and "buccal swabs".) The DNA samples are then processed by the laboratory to compare the DNA of the patients.

No two people have exactly the same DNA (with the exception of identical twins).

We get half of our DNA from our father, and the other half from our mother. When we do a DNA paternity test, we compare the DNA of the child to the father. For example, the top 3 lines (out of many) of a sample DNA Test Report showing a child getting half their DNA from the mother and half from the father:

By comparing a large number of locations (loci, plural of locus) in DNA samples, we can tell with accuracy of 99.99+ percent whether or not the child comes from the father. The result is normally either 99.99+% or else 0%.

However, we must choose carefully which locations to compare, because the DNA between any two humans is about 99.9% the same (and humans are over 95% the same as some other primates such as monkeys, with bonobos the closest at around 98.7% the same as humans, and chimpanzees nearly the same amount). Therefore, between humans, we must know where on the other 0.1% to compare the DNA. Within this 0.1% are the variations from person to person such as hair color, eye color, skin color, and lots of other things.

Of this 0.1%, some of that has a little variation between people, and other parts have a lot of variation. We must be careful which parts of the DNA we compare. We compare a much smaller fraction where there is much more variation in the human population. We also choose locations ("loci") for which there are very large statistics of samples from people in scientific studies.

Our laboratory is one of the oldest and biggest laboratories in the world, whereby our laboratory has privately built up its own database of human DNA variations over decades, and its statistics on the human population. This database is used to calculate the % probability of paternity.

Anybody can say 99.9999999%, but is it true? We show our calculation in detail.

For a given test, the result is normally 99.99+ percent, or 0%. It is very clear, father or not father.

On each person's DNA, we check at least 20 standard locations ("loci", plural for "locus"). We measure the length of the DNA "allele" at each locus. (There is also one locus for gender, which just tells whether the person is a male (XY) or female (X), since only males have a Y chromosome.)

(Our laboratory previously offered a minimum of 16 loci, many years ago, but that was increased to at least 20 at the same price, though some of the text and graphics on this website is from the time when 16 was offered.)

When we look at the laboratory report, we see two columns for each person, each with a number value equal to the size of DNA allele at that "locus" location, the two alleles of the child, the two alleles of the alleged father, and optionally the two alleles of the mother if she was also tested. The table normally lists at least 20 locus rows. At each locus we look at the child's two values (two alleles) first. One of the DNA allele values must come from the mother, and the other one must come from the father, in a paternity test. For example, if a child's alleles have the two values of 14 and 18, then the mother will have one of those two values in common (for example, she may have 5 and 14). If the mother has 14 in common then the father must have an 18 as one of his two values if he is the biological father. If he does not, then he is excluded -- not the father. (See rare exceptions discussed later, a mutation.) However, if he does have an 18, then that row (locus) has a match.

For a test of 20 loci (rows), comparing any two people (for example, comparing you with me, whereby we are not related) will probably have matches for several loci but for sure will have several mismatches for other loci. In a DNA test where the man is not the father, there will typically be 5 or more points of mismatch yet 5 or more matches, too. The matches are simply because all humans share over 99.9% of our DNA, and of the 0.1% remainder, different ancestors and families still share various alleles.

However, in a DNA test where the man is the father, there will normally be NO mismatches at all in these loci (see rare exceptions discussed later, a mutation). When a calculation is done, the probability of paternity will normally be over 99.99% or else 0%.

There are two testing situations:

1. "Trio" -- 3 people -- mother, child, and alleged father are all tested
   
2. "Duo" -- 2 people -- only child and alleged father, no sample from the mother

For a Trio, for each locus (line), we first identify which one allele (column) they got from their mother, meaning that the other allele must come from the father, which we look for second. Of the father's two alleles, either of those two must match the child's one allele unclaimed by the mother's DNA. (The father and mother can pass on only one of their two alleles, and it can be either.)

For a Duo, with no mother's sample, we can compare only the father's alleles to the child's alleles, so that one of the two alleles must come from the father, for each locus -- either of the father's two alleles should match either of the child's two alleles, for every locus (see rare exceptions below, a possible "mutation").

The results are extremely accurate, for either a trio or a duo. If the man is the father, then the probability of paternity will be greater than 99.99%. If the man is not the father, then the probability of paternity will be = 0%. The test is thorough enough to be sure it is either 0% or else a minimum of 99.99% accurate, whether a duo or a trio.

There are two main benefits of a trio: 1. You can see that the child is truly from the mother, because there will normally be a match on every line for the mother, whether or not there is a match for the father. (There have been cases where the child was not from the alleged mother, plus not from the alleged father, of course. Those are unusual stories...) 2. The probability of paternity will be higher for a trio than a duo, if the man is the father, for example, 99.9999999% (and often even higher than that), though a duo is normally still at least 99.99% if the man is the father. If the man is not the father, it's still 0% for both trio and duo.

When the man is the father, the percent probability is not the same for every man. It varies from case to case depending upon one's particular alleles -- how common or unusual the matched alleles are within the human population. For example, if you have some rare DNA, and the child has this exact same rare DNA, then your percent probability will be higher than for a man who shares common DNA with his child. In any case, even for men with relatively common DNA, the percent probability will still be more than 99.99%

In scientific language, the report will not say you "are the father", it will just state a probability of paternity which is usually 99.99...%, or else it will say 0% and that you are "excluded".

When a man is excluded, there are usually at least 5 points of difference, and often many more than that. It is very clear. There are usually at least 5 points that match simply because humans are similar in DNA. It is rare that there are less than 5 points of mismatch in a 20 loci test.

However, if there is only one mismatch, then we do not jump to the conclusion that the man cannot be the father. Occasionally there is a "mutation" whereby there can be one mismatch. It is extremely rare to have 2 mismatches. Therefore, we need at least three mismatches to "exclude" a man as biological father.

In cases of one mismatch, the % probability of paternity can be lower than the threshold of 99.99% after testing 20 loci, so we can continue testing additional loci beyond the standard 20, such as 25 or more, until the probability is over 99.99% (or else 0%).

Notably, we have never yet had any case where it was not clear whether or not the man was the father. For the first 20 loci, we have not yet seen a result with just 2, 3, or 4 mismatches. It has been either 0, 1, or else 5+ mismatches. We have had many cases over the years with 1 mutation. When there was only 1 mismatch out of 20 loci, we've never had a father excluded to date by testing additional loci beyond 20. It has just been a matter of testing additional loci until the probability is above 99.99%. However, in many cases, the percent probability is already over 99.99% with just 20 loci even when one is a mutation. The percent probability depends upon how common your alleles are in the human population for the loci tested.

Based on the above explanation, a sample report for a trio looks like the one below. (For the blanks in the report, see the explanation below this graphic.)

The purple circles denote the first 4 matches to the mother.
The green circles denote the first 4 matches to the alleged father.
You can match the other 11 alleles as an exercise.

If no sample was submitted for the mother, i.e., if it was a duo, then there would be only 2 columns instead of 3, i.e., there would be no column for the mother.

The bottom loci, "Amelogenin", reveals the gender of each individual. X = Female. XY = Male.

In nearly all cases, there are some blanks on the sheet, whereby an individual has only one allele number reported on a particular row (loci), instead of two numbers. Here is why:

It is common for a person's two allele numbers to be identical, for example 12 and 12, instead of having two different numbers, for example 13 and 16. If the numbers are identical, for example 12 and 12, then in order to be compliant with the main American accrediting agency (AABB), the testing laboratory is permitted to report only one number. (When both alleles are the same for one locus, it is called a "homozygote".)

A more detailed explanation is that, in the laboratory, for each locus, the two alleles overlap in the readout, as shown on the right side below. For example, in the sample on the left side below, you can see two clear peaks, at 13 and 16. However, if the two allele sizes are identical, then they would overlap, which would show up as only one peak, as in the example on the right side. For a reading like the one on the right, it is usually considered a safe interpretation that it is a 12, 12 (homozygote),

However, in order to be compliant with AABB standards, testing laboratories are permitted to report only what is observed, not what is assumed or interpreted. The guidelines state "For any apparent homozygote, only the observed phenotype shall be listed. If a single band is visualized, its measured band size or repeat number is to be reported only once".

Reports from other laboratories which are not AABB compliant often list the results as  " 12   12 " which may be clearer to the ordinary person reading the report, but there is actually no observation of two 12's in the laboratory, only one 12.

(However, a homozygote should not always be an automatic conclusion, because of possible "allele dropout", i.e., an allele can possibly exist which is not detected and thus not reported. While very rare for paternity tests on the selected loci using standard equipment for this purpose, generally in the much greater DNA analysis world there exists allele dropout in testing procedures, for various reasons. This seems to be very rare in the paternity testing realm using standard kit, but much more common in wider general research. This is one reason it's better to not jump to a " 12 12 " conclusion, and instead report only what was observed, which is safer when interpreting a DNA test overall. It's also a good reason to stick with a laboratory which does paternity testing on well known loci using well understood testing procedures, good kit, and professional staff who are careful what they report.)

How the probability of paternity is calculated

On the left side of the report, there is a column called "PI", which is "Paternity Index". Each loci has a different PI. The value of the PI will vary from report to report, depending upon a person's DNA.

For example, if a child and an alleged father match on an allele size of 22, but an allele size of 22 is very common in the human population, such as 40% of people in the world having an allele size of 22, then the PI will be low, meaning that this does not give a high probability of paternity for that one allele locus. However, if they match on an allele size of 8, and only about 5% of people have an allele size of 8, then the PI will be high for that one allele locus, meaning it gives a higher probability of paternity.

If you multiply together all the PI numbers, then you get the number at the bottom, the "Combined Paternity Index" CPI, for example, 323,769. What this means is that if we did a DNA test of every man, only about one man out of 323,769 men would "pass" a paternity test for this child, and the other 323,768 would fail it, generally speaking.

If only 1 out of 323,769 men would pass the test, then only
00.0004% would pass the test, so that the probability of paternity is
100.0000 - 00.0004 = 99.9996%, which is shown on the rightside.

(Actually, if you multiply all the PI numbers, you will probably get a very slightly different number for the CPI, because the actual PI may be 1.548 but the report shows only the rounded 1.55 for brevity, though the calculated number uses the slightly more accurate inputs.)

Most courts and other purposes require a CPI of at least 200 or a probability of paternity of 99.5% or better. Many now require 99.99% minimum as standard. Our laboratory reports a minimum of 99.99% for paternity or maternity inclusion, or else 0%. For example, if there is one or more possible mutations in the first 20 loci tested which drops the probability of paternity to below 99.99% but still above 0%, then the laboratory would continue to test additional loci beyond the first 20 until the conclusion is either 0% or at least 99.99% probability. The laboratory has the capability to test a very large number of loci.

Calculating all the PI numbers and the CPI requires professional expertise. Some laboratories give you just a CPI and probability of paternity without showing you how they calculated it. Our reports show you each and every PI number for each and every locus.

We've had customers who tested with a second laboratory and compared their results to ours, and showed us the two reports side by side. Some laboratories have stated a higher % probability of paternity with more 9's, e.g., 99.999999% instead of 99.996%, and locus-by-locus PI numbers differed even though it was the exact same allele data for those loci. Obviously, one or the other is not really accurate. Both are saying a very high probability that the alleged father is the biological father, and both percentages may be good enough for the client, but one service provider may be able to brag about their higher percentage, whether or not it's really accurate. It might possibly be an honest calculation but based on other population data.

Thus, just because one laboratory says 99.999999% and another says 99.998%, it doesn't mean that the first laboratory is better than the second laboratory. Which calculation is really more accurate?

For example, a laboratory which tests mainly Americans and has much less Asian data might come up with a higher percentage probability of paternity for a case involving Southeast Asians, if some of the Southeast Asian DNA profile data may be a bit unusual compared to American people in an American database. The more unusual an allele match, the higher the probability of paternity can be in a calculation.

(It is also possible to look at many personal profiles and come up with a probable racial profile, and a probability of what general part(s) of the world their ancestors probably came from.)

Some calculations can be based on reported race on the sample, or apparent race derived from the DNA profile data, which can improve the probability calculation, compared to just a simple calculation.

All laboratories don't use the same population data in their calculations. It's not as if they all tap into some central world database. Many laboratories and other sources don't share their population data with others, and may keep it confidential. A particular laboratory can get their data wherever they can find it available. It can be based on their past customers, and/or it can be taken in from various other sources which might have regional biases in them.

It's notable that our laboratory was an early DNA testing facility, is very large, and has long used international partners. It has built up a vast database of DNA allele frequencies in the human population on which to base its calculations.

The above explanation is enough to understand a DNA Test Report. The little bit below is an extra for curious people.

When the DNA samples are received and processed, the first step is to strip away everything else in the cell except the DNA, and get a good initial DNA sample to process further.

After that, DNA is then replicated multiple times or "amplified" by a process called "polymerase chain reaction" (PCR) to get a much larger sample to test. Since COVID-19 came out, people have heard about "PCR" tests, which take much longer than antigen tests, but are more sensitive and accurate. The PCR process has long been used in DNA profiling, and actually goes back to the 1980s. In the natural world, cells replicate their DNA when you grow and during body maintenance, and this is basically the same way we do it artificially in the laboratory. (When a cell divides in your body, enzymes called polymerases do the copying.)

This cycle of "DNA amplification" can be repeated as many times as necessary to provide enough DNA material from the specimen to get a strong enough readout in the laboratory.

We have tried to make this page as easy as possible to understand for the majority of our customers in Thailand. Please let us know if there is anything you don't understand, or anything further you would like to know.

Homepage

Paternity DNA Test

How DNA Testing Works

Prices

Remote Locations

Mailed kits, self-collection

No Blood, No Pain

Anonymity and confidentiality

Personal vs. Legal Tests

Western Embassies, Immigration, Thai Government

Corruption and Bribes

Can you trust us?

Psychological Issues

Official Certifications

Hair, blood, fingernails...

Require Breastfeeding

Other DNA Tests

Policies and Confidentiality

FAQ

About Us

Contact Us, Map

    Menu: see our Site Map
    This site is about
    Thailand DNA testing.