Saturday, May 22, 2010

What do my siblingship DNA test results mean?

One of the most common searches that people find my blog through is siblingship DNA tests. In the past few months I have received MANY emails from adoptees and donor-conceived adults who have gone through a DNA test with a possible sibling - only to receive results they don't really understand. Some simply do not understand what their results mean. Others are confused as to why they are inconclusive. Many others think that there's still a possibly of them being siblings when in reality the results are VERY negative.

Most just want the peace of mind of knowing if they are siblings or not. For most of these people I am able to give them some concrete answers or lead them in a direction to provide them better results.

This shows me that most DNA testing companies are doing a less than satisfactory job at explaining what an individual's results mean in plain language --- a catch that many in the medical field struggle with. Medical information to the layman means nothing if they cannot comprehend what they are being told or reading.

So I thought, rather than talking about DNA testing in general or what all the different numbers mean, I wanted to discuss negative and inconclusive results as well as postives and provide some examples of these types of results as well as how DNA testing companies word results (that make them indecipherable to those without background knowledge).

Below are some examples of actual DNA tests (names have been removed to protect the individuals) with some explanation as to what they mean.


Example #1: Conclusive sibling match w/ both mothers tested.

Below are results of a siblingship DNA test with two alleged siblings and both their mothers that is clearly a conclusive match.


In the above example the two alleged siblings and both their mothers were tested. When both mothers are tested it allows the analyzers to determine which allele at a specific marker came from the alleged shared biological father. If the two alleged siblings share this allele it increases the individual SI for that marker. The results are in the yellow box below. This is very obviously a sibling match as they are 99.8% probability of being siblings. The CSI (combined siblingship index) is over 500, signifying a very strong conclusive result. No questions asked, these two individuals are most definitely half-siblings.

However, as I show below, even in the case where two individuals are most definitely siblings in one test, take away the mothers and those results change dramatically!!


Example #2: Inconclusive sibling test w/ neither mother tested.

If we took the same two siblings in Example #1, but neither of their moms were tested, here's what we would have...


The same two individuals that in Example #1 were surely siblings, these results are questionable, and would be considered inconclusive. They cannot be deemed negative because they are between 0.5 and 1.0 (although nearly on the brink of going below 0.5). Their probability is only at 33% but their odds are 1.95 to 1. That is....they are 1.95 times more likely of NOT being half-siblings than being half-siblings. These odds, while still suggesting inconclusive negative results, are still fairly low.

Odds that are more like 20 to 1, 50 to 1, 100 to 1, are more significantly negative results. A result received like in Example #2 suggests to me that further testing should be done.....especially if one or both mothers were not tested!!!! If both mothers ARE tested, I would suggest a second panel of STR markers (autosomal markers like the ones above...most DNA testing companies can do extra if you request them) or either X or Y chromosome testing (if the two alleged siblings are of the same sex). The latter is preferable for same-sex alleged sibling pairs.


Example #3: Inconclusive sibling test w/ one mother tested.

Here is another example of an inconclusive result. In this case only one mother has been tested (as can be often the case with older offspring and adoptees where one mother may have already passed away).


The results suggest relatedness, as in the CSI is over 1.0 and the probability is over 50%, however it is only over the threshold by a smidgen.

The two examples I showed above provide reason as to why I STRONGLY suggest at least one mother to be tested!!! I know there are many siblingship tests that are found to be inconclusive or downright negative when in fact they may actually be siblings, simply because neither mother is tested. In Example #2, these two individuals, if one of the mothers were tested, their results would have been at 92% probability - enough for the US Courts, but would not have been successful with UKDL's stringent 99% cut-off. In Example #3, had it been possible (one mother was deceased), a further test taking into account Sibling 2's mother could possibly have provided these two individuals the peace of mind of knowing that they were or were not truly siblings. I suggested further STR marker tests for these two.

It's this probability and odds that confuse many people, and are what constitute the majority of questions I receive. Remember: Siblingship tests are based on a prior probability of 50%. This means that 50% probability is the starting point and and results that lean towards being positive will raise that probability above 50% and results that lean towards being negative will lower that probability below 50%.

Some people are confused and this that a probability of say 20% means the probability of two individuals being siblings picked from random - i.e. 20% with zero as the starting point. This makes a 20% probability sound relatively supportive. However, when a siblingship test is being conducted, they are already under the assumption of being possibly related and not random individuals, thus the prior probability is 50% (there's a 50-50 chance that they are or are not related). Therefore, a probability of 20% is NOT good.


Example #4: Conclusive negative sibling test w/ both mothers tested.

Below is a conclusive negative test, with not questions asked these two individuals are not related at all.


In this example these two individuals are conclusively not siblings. With results like these it would be nearly impossible for further testing to provide anything but similar negative results. The combined siblingship index is waaaayyyy below 1.0, and the odds are over 1,000.


I hope these examples and their explanations have helped many of my readers and those that find my blog for these questions, to better understand your results, and have more confidence in what they mean. As always, please email me if you have any further questions or would like me to take a look at any results and provide a better understanding into what they mean.

17 comments:

Anonymous said...

Nice explanation. I am curious about how the prior probability of 50% was arrived at. It seems to me, naively (I am NOT a mathematician) that before you have any data the chances of any two people being siblings would be much lower (this is how I understood prior probabilities to work...but I am prepared to be wrong)

Lindsay said...

Prior probability is a really complicated concept to understand, but it basically is a way of determining what the probability of a specific event occurring before any data is collected.

Siblingship and paternity are examples that have two results only (this does not mean DNA test results, but rather true answers). Either two individuals can or they cannot be siblings, or parent and child. There is no in-between (but there can be on a DNA test, but that is a fault with the test and not with the relationship between those two individuals).

In our case it is the probability of the event of two individuals being siblings. So if you take sibling 1 and any other untested individual, there would be a 50% chance that the second individual is a sibling and a 50% chance that they are not a sibling --- since they either can or cannot be.

Then, after the testing is through whatever the probability is, it must be compared back to that prior probability.

What you are talking about is ANY two individuals. We are talking about two specific individuals. Again, I know that's kind of confusing, but you're right, the chance of any two people being siblings is very low, which is why siblingship tests can be done. But once you pick two random individuals to be tested that probability becomes 50% because either they are or they are not siblings.

So when two individuals get a result that says 20% probability of relatedness, that is why that is a bad probability. It might seem high (if we were comparing ANY two individuals) but for these two specific individuals with a prior probability of 50% that is not a good score.

Make any more sense to you? Sorry, I know this was not the best explanation of prior probability but I'm really not sure the best way to describe it.

Anonymous said...

OK..so I have a supposed half brother and we want to be tested. My mother is willing to test but his mother isn't. Our father is deceased, cremated and his ashes are scattered in the ocean..so we can't obviously test him. With only 3 to test..how conclusive will our test be and will it be submisable in court??

Lindsay said...

Anonymous,
The thing with siblingship, and especially half-siblingship, is that you can't tell how "conclusive" the results are until the test is done. The good news is that with one of your mothers tested it greatly increases the conclusiveness of the results....meaning it will push that siblingship index either higher above 1.0 if you are true half-siblings, or well-below 1.0 if you are not.

I can't tell you for sure if it will be conclusive or not, that depends on the results. But as for court submissable, if you plan to use this test result in court you MUST get a legally-binding test which requires witnesses to be present during the test or go to a special location. If you look at different DNA testing companies you will see the costs, but for example Genetics Testing Laboratories has a legally-binding siblingship test, with one parent, for $350.

Hope that helps.

Anonymous said...

My supposedly Half-Brother and I had a DNA test.I found him through searching as I was adopted.We think we have the same father. The wording of the results is confusing.Prior Probability is 50%. We have Seven out of 15 matches, but the wording says 2.33 less likely verses no relation.It sounds negative.Are we Half-Siblings or not!

Lindsay said...

Hi Anonymous,

What prior probability means is that before the test is done there is a 50% chance that two individuals are related or not related. Once the test is done the results either are greater than 50% (suggesting relatedness) or less than 50% (suggesting unrelatedness). The higher or lower that probability is away from 50% (over 90% or below 10%), the more conclusive the results.

Now, your results say that you're 2.33 times less likely of being related that means your "likelihood ratio" is around 0.42 and your probability of relatedness is about 19%. But remember, the prior probability is 50% so 19% is quite a bit lower than 50% which suggests a negative result.

However, if neither of your mothers were tested it's possible that it's a false negative result simply because it's unable to prove if the shared alleles are from the same alleged parent. I would suggest (if it's possible) to have at least one of your mothers tested...I don't know if you have found your b-mom or not. But even one mother tested would greatly increase the accuracy of the test.

It's also possible/likely that the alleles you do share are all very common in the general population, thus without any mothers DNA tested it's almost impossible to prove siblingship without a test such as a Y-STR test or an autosomal SNP test like FamilyTreeDNA's Family Finder test. I would strongly suggest if you want CONCLUSIVE proof if you are half-brothers or not to look into either a 37 or 67 marker Y-DNA test.

Susan Patterson said...

I'm the writer from yesterday! I called the DNA people to explain the results and they said David and I have a 30% chance with 50% being the baseline so not conclusive, but not excluded either. We didn't have any parents included as they're all deceased.So, I guess we'll assume we are as some info matches from my adoption . Thanks for replying. Susan

Lindsay said...

Hi Susan,

Yes, it's inconclusive, but without any parents tested there can't be any more accuracy. I would strongly suggest doing FamilyTreeDNA's Family Finder test, as it can identify with almost 100% accuracy (far better than a siblingship test) if two individuals are half-siblings or not.

Also, as an adoptee it's a great option for learning more about your b-parents families. And if this suspected half-brother turns out to not be related, then the test may help you identify information about your b-father.


Lindsay

Anonymous said...

hi lindsay could you help me understand my daughters dna results if i email them please? my email is welshdragon2@hotmail.com

Anonymous said...

Hi Lindsay,

First of all, I want to thank you for taking the time to help people like me who are confused and frustrated when trying to interpret siblingship DNA test results. I just received my DNA Sibship Results and am confused as to the findings: COMBINED SIBSHIP INDEX (FULL SIBLING): 12.83; COMBINED SIBSHIP INDEX (HALF-SIBLING): 27.60. I had submitted buccal swab samples for my sister and myself. I know we share the same mother, but I am doubtful we share the same father, who is deceased. Can you please explain the results to me as to the probability that we have different fathers? I know you understand the importance of this information to me and would be extremely grateful for your help.

Anonymous said...

Good evening Lindsay,
I have emailed you my results and I would really appreciate it if you looked at them for me and tell me more about the probability index. They have ruled out our relationship fromwhat i can gather on the interpretation.
How do you work the probability index or combined one out? Is it that the more matches you have you are less likely to be biologically related or the opposite? Do they have to be different?
Would you also be able whether my company I used have done a 37 or 67 Y chromosone test but I am not sure how you could tell that?

Thank you
Besnik

Sue Ellen Howard said...

Hi Lindsay,
I want to begin by saying that I love your site and thank you for making this information available! I was adopted at birth and found my birthmother and three older siblings 25 years ago. According to my mother, she would maintain a relationship with me as long as I didn't tell one of my brothers that I was his sister as she didn't want his father to know about me. She refuses to tell me who my father is and stated that she'd tell me when he father of her other children passes away. Well, he's healthy, she's not, and I'll never know now. Long story short, I have contacted my brother after all of these years, taken a DNA siblingship test with him (we are wondering if I share the same father), and we still don't know. No parents were tested with us, just one brother (him) and one sister (me) The results came back super high on half siblingship (19.43=95.1%) and super low on the full siblinghip (.31 or 23.6%). His father is willing to take a DNA test with us if this is considered inconclusive, but if leaves no question, then we'll save our money and call it good. I haven't seen any examples with results skewed so in opposite directions like that. Can you please help me? Thank you SO much! Sincerely,
Sue Ellen
ps-I had 5 dashes with no numbers on my test, and he had 2 on his...what does that mean? They tested 16 alelles, but I only saw 15.

Anonymous said...

hello Lindsay, i had a sibling test done for my two children one being a boy and the other being a girl. i had the test done so i could know if they shared the same father, fulling knowing im the mother but i did not get tested. the results came back as 99.99% that are mor likely full biological siblings versus not being related at all. it then said in the letter, these results do not supercede testing involving additional participants related to my son and daughter. inclusion of the known parent(s) may lead additional results to this relationship and is highly recommended if at all possible. I am unsure if this means that could possibly have different fathers or if i got tested myself could this then change the results so they are half siblings? please could you put my mind at rest as i can not get the father tested as he has nothing to do with either children many thanks

Anonymous said...

Just received the results of a paternity test which revealed the "father" was excluded as the bio-father. It is believed that my potential bio-sibs are now half. Apparently, a paternity test had been done fairly recently with the son of the same father. In that case, the father was not excluded as the bio-father. Do you know if it's possible to examine the reports of both "children" where CODIS markers are listed on each report to assist in determining whether we share the same bio-mother? Bio-mother is deceased, and I am a female adoptee.

Anonymous said...

Hi Lindsay,
If I register my adopted Daughter from China DNA with Family Tree DNA or 23 and Me (without a sample from bio Mom or Dad) and they do the y-STR and or autosmal SNP test would they be able to Identify another full sibling who was also registered with the co.? Could they conclude with 100% certainty and if not 100% what would the range be?
Thank you! Ann

Anonymous said...

Hi Lindsey,
I just got my sibling results for my daughter and it said 1.84 times (64.75%) that they are full siblings. I was the only parent tested.is that a high score or dos that mean its a slight chance their full sibling.Thank you for your help.

JD said...

Hi Lindsay,

I work for a DNA testing company and let me say what a GREAT site you have. While we do spend a great deal of time explaining the testing results on sibling tests, I also think your site would be great to refer people to for their own studies.

One of the big challenges faced with sibling testing is performing testing using a limited number of markers. When we do testing we prefer to test using a 41 marker basis. Of course it is great if the mother is available, however, using a higher marker base is able to give more accurate results in most cases even if the mother does not test.

Keep up the good work! And thanks for what you do.