How accurate are dna tests for half-siblingship?

Question

If a dna test for half-siblingship (where both Mother's were tested also) reports 6% chance. How reliable is this?

Both mothers were tested against both children, the the remaining dna was tested between the two children to determine if they share the same father (who is not available for testing). 9 out of 15 alleles matched. They concluded that the probability of a half-sibling relationship was 6%.

Answer 1

Where you have missing data, in this case the father, then there is always a higher risk of error, than with a complete dataset.

Here are a few potential sources of error:

First of all you have to constrain the mother offspring comparisons such that there are no mismatches - what I mean is that in the case where the mother has one allele that is also present in the offspring then it is assumed that the offspring inherited that from its mother. However, with this type of hypervariable locus it does happen that new alleles are generated more frequently than with other types of 'gene'. There is no way of knowing whether there are any mismatches in this case, so this is one potential source of error, even if the chances are quite small.

The next problem is as follows: assume mother has genotype AB at one locus and so does the offspring, then the mother could have passed on either allele. Obviously, the statistical models they run will calculate probabilities based on both scenarios and the best fit with data from other loci will be chosen, but again it could be wrong.

Given that all of the above are incorporated into the model that calculates the probabilities, then the final step is to combine the remaining alleles from the two offspring to create a multi-locus genotype for the potential father. Where offspring share an allele then only half a gentoype is present for the other six loci a complete genotype is generated.

The probability of the father's genotype will be prone to more error as some alleles are missing, so there are only 6 complete data points, so there is the potential for statistical error on such a low sample size.

There is one final issue relating to this - the probabilities are based on comparison with a population. How well the probability of an individual's genotype can be calculated on how well he matches the test population. If he is a white man from London then probably the test will be reasonably reliable, if he is a Native Indian then they will be useless as the distribution of alleles will be completely different.

There are some sources of error, but do I believe the outcome? I would like to see such results given with some estimate of the error, which any scientist would have to give presenting statistical results. If the error were 0.5% then I would be more inclined to accept the results than if the error were 10%, it is the error value you really need to know.

Sorry if it's a bit waffly, it's a bit complicated and there is not enough time or space here.

Answer 2

Your question is a bit unclear. It sounds like the 6% chance is trying to answer your question. I am guessing 2 kids and 2 mothers were available for testing but the possible father in common was not available. right? I would imagine that the test could be extremely accurate depending on what is done in the lab and what is found in the test...
can you quote the part of the test that says 6%?

Answer 3

The DNA tests themselves are very, very accurate.

The siblings matched 9 out of 15 allelles? It depends of the exact allelles that were used, but forensically 9/15 allelles is common is enough to assume a sibling relationship...

Also, with 9/15 in common, there is no way the result would only be 6%. I've attached a portion of one of my assignments (the whole thing is 19 pages) in which I showed statistically that even one, very common allele in a paternity case would give you at least a 64% chance...
____________________________________________
Daughter’s Alleles0.5/Allele Freq.1.0/Allele Freq.
D16S53914, 1528.2515, 181.77 - 1.77
Combined Paternity Index 1.77

Theoretical Probability of Paternity: (1.77 / (1+1.77))*100= 63.899%
__________________________________________


But then again, that's an excellent of example of why I don't put much stock in the use of statistics for anything smaller than a population...

Answer 4

i'm in the medical field DNA exams attempt DNA Markers between the guy being examined, the newborn and the mummy of the newborn.. they're ninety 9.9% precise to quote Maury Povich Your boyfriend is "no longer the daddy" and she is both mendacity about who she slept with, she replaced into below the impression or she may were drugged and raped and that's the reason she would not bear in options....

Answer 5

DNA testing is quite reliable. i believe about 99.99% accurate.

In these tests what they do is test for patterns in your DNA and compare them to your parents'. Any pattern in the child's DNA would be found in their father's or mother's DNA.

Answer 6

100% accurate, subject to test done by an experienced person, in a well equipped Lab.