Saturday, November 3, 2012

Autosomal DNA Testing: Recombination

     Good Day Everyone. How is everyone doing???  Fine I hope. In this document, we are going to review the natural and biological process of recombination. Recombination is an important concept that one should probably be famaliar with in DNA Genealogy. This is especially true in Autosomal DNA testing. The other DNA tests, Y-DNA and mtDNA, are immune to the effects of recombination. The same can not be said for an autosomal DNA test. An autosomal DNA test is constrained and effected by genetic recombination. In this document, we are going to see why. Let's begin with taking a high level look at recombination. Then, we will move to a more low level view of recombination and see how the tests are affected. Let's begin.

High Level View: Basics Of Genetic Recombination 
Genetic recombination is defined as a biological process where genetic material is broken and joined to form new genetic material. The diversity of life that surrounds you is due to recombination. Recombination occurs when a child is being conceived for the first time. Humans have 46 chromosomes. When people have children, only 23 chromsomes (half) are passed to the child from each parent. The key is that 23 new chromosomes from each parent are created and passed to the child. Here is what is meant by new.

Basic Mechanism of Recombination In A Parent
Start                              Touch                      New Chromosomes
C(Blue)--><--C(Red)     C(Blue)C(Red)         C(Blue/Red)<--->C(Red/Blue)

Let's start off with two chromosomes which we'll call -> C. One chromosome is blue -> C(Blue) and the other chromosome is red -> C(Red). In each parent when conception occurs, the 46 chromosomes divide up into 23 pairs. Each pair consists of two chromosomes like shown above. What happens in a parent is that both chromosomes within a pair, physically touch each other and separate. When the two chromosomes touch each other, they exchange genetic information. The result is that after separation, two new chromosomes are created. 23 new chromosomes are then placed into a sperm cell and 23 new chromosomes are placed into an egg cell -> sperm[23,C(Blue/Red)] and egg[23,C(Red/Blue)].

When a sperm and egg cell combine to form a new child, the child will have 46 new chromsomes. This is why siblings (brothers and sisters) look different from the each other and from the parents. In a nut shell, that's basically what recombination entails. The exception to this is identical twins. Without recombination, this would be a boring world as everyone would look the same.

Now let's look at recombination from a low level view.

Low Level View: Exchange of Genetic Material
DNA is composed of four bases -> A, T, C, and G. A DNA segment would look like this -> ATTTTCGC. Let's take a look at recombination at the DNA segment level.

Start                                                                   Touch
Chr1-AAAAAA->  <-TTTTTT-Chr2                     Chr1-AAAAAATTTTTT-Chr2  

New Segments

Shown above is an example of recombination at the DNA level. We start of with two chromosomes -> Chr1 and Chr2. Basically recombination has created two new DNA segments -> AAATTT and AAATTT. The two DNA segments have exchanged DNA or genetic material with each other. Of course there are many possible combinations that can be generated from recombination. We could of gotten ATAAAA, TTAATA, etc.

Genetic recombination is responsible for another process, the loss of genetic material.

Low Level View: Loss of DNA
Each of us has 46 chromosomes. This is a basic fact. 23 of our 46 chromosomes we inherit from mom, and the other 23 we get from dad. In other words, 50 percent (23/46), or half of your DNA, is from each parent. If 50 percent, or half of your DNA, is from your mom, then it stands to reason that half of that 50 percent, or 25 percent, is from your mother's parents. In other words, we each get 25 percent of DNA from our grandparents.

This is what basically happens over time

50% parent -> 25% grandparent -> 12.5% great-grandparent -> 6.2% great-great grandparents -> etc.

As you can see, the amount or percentage of DNA that's inherited from an ancestor gets smaller as you go back further into the past. Each generation, the percentage of your previous ancestor's contribution to your DNA is halved. The reason for this is recombination. What this means is that a DNA segment from an ancestor gets smaller and smaller as the DNA segment is passed down thru the generations.

Ancestor-AAAATTTTGGGGCC --> Child-AAAATTTTGGGG --> Grandchild-AAAA --> Great Grandchild-AAA --> etc

Shown above is an example of recombination at work over the generations. We start off with a DNA segment in a ancestor that has 15 DNA bases. Over time, this DNA segment has been reduced as it's passed down through the generations. Notice that as the DNA segment was passed from the child to the grandchild, a large chunk of that DNA segment was removed!!!!! On the other hand, notice from grandchild to great-grandchild, only a single DNA base was removed. This highlights an important theme when dealing with recombination.

IMPORTANT: The amount of DNA that is lost between generations is random.

This is by far the most important concept to grasp and understand. Recombination is an unbias and haphazard process. Recombination doesn't care. It removes DNA in a completely random, unbias, and unpredictable pattern and fashion. The amount removed can be quite large, or quite small. It's makes no difference.

Now let's look at autosomal DNA tests in light of recombination.

Recombination: Autosomal DNA Testing 
Now that you are armed with the basics of recombination, we can look at how an autosomal DNA test operates. Behind the scenes of an autosomal DNA test such as Family Finder or Relative Finder is a matching algorithm that performs the work. The matching algorithm identifies linked DNA segments that would be present in two or more people whom are descended from a common ancestor. These linked DNA segments are composed of DNA bases called SNPs (snips). If there are enough matching DNA segments between two or more people, a "match" is declared.

     If two or more people are descended from a common ancestor, they both should share linked DNA segments from that common ancestor. Simple enough right??? However there is one player in the game that needs to be recognized -> recombination. As those segments are passed down across the generations, recombination will shorten those linked DNA segments. In some cases, recombination may even completely "erase" some of those linked DNA segments. That's why an autosomal DNA test can only go back 5 to 7 generations. After a certain period of time (5 or 7 generations), recombination has the potential to completely erase linked DNA segments from a shared common ancestor.

Let's take a look
Line 1: Ancestor-AAAATTTTGGGGCC --> Child-AAAATTTTGGGG -->  Grandchild-AAAATTT --> Great Grandchild-AAAATT ---> person A (match)

Line 2: Ancestor-AAAATTTTGGGGCC --> Child-AAAATTTTGG -->   Grandchild-AAAATTTTG --> Great Grandchild-AAAATTG --> person B (match)

Line 3: Ancestor-AAAATTTTGGGGCC --> Child-AAAAT -->   Grandchild-AAA -->
Great Grandchild-AA -->  person C  (non-match)

Shown above are three lines of descent from a shared common ancestor in three people A, B, and C. Let's assume we know before hand, that these people are related and are descended from the same ancestor. The matching algorithm doesn't know the three people are related. The matching algorithm only knows that its job is to look for matching DNA segments. Let's say in this example, that the matching algorithm will declare a match if the DNA linked segments are identical and have at least six matching SNPs. In this case, the algorithm will declare both person A-->AAAATT and person B- ->AAAATTG as a match.

Person C will be left out. As we can see, recombination has randomly removed enough SNPs from Person C's DNA segment such that the matching algorithm will not declare a match. This highlights an important theme in autosomal DNA testing.

IMPORTANT -> Just because two or more people don't match, doesn't necessarily mean they aren't related in genealogical time.

Now let's look at some examples of recombination and it's effects

Example 1: Loss Of Large cM Amounts
Here is an example from my personal family. Remember that the centiMorgan (cM) is the unit of measurement that is relevant in an autosomal DNA test. The centiMorgan contains various DNA testing properties such as DNA segment length, number of SNPs, etc, all rolled into one. The centiMorgan gives us a consistent way to compare apples to apples and make judgements as to whether two or people are related.

Technically, the centiMorgan gives a probability or propensity of recombination. But let's keep things simple.

Family Case 1: Match A
Juliette Turner (grandmother) & Match A
Shared cM -> 65.13
Longest segment -> 42.96cM

Steve Handy Sr (father) & Match A
Shared cM -> 62.02
Longest Segment -> 42.96cM

Steve Handy Jr (myself via 23andME) & Match A
Shared cM -> 12
Longest Segment -> 12cM

As one can see, the effects of recombination are truly random and unbiased. In the generation between my grandmother and dad, only 3.11cMs of DNA were lost (65.13cM - 62.02cM). However in the generation between my dad and myself, over 50cMs were lost!!! (62.02cM - 12cM). This is how recombination operates. It operates in a haphazard and unpredictable manner. In essence, a matching algorithm deals with the results after recombination has done it's part.

Now let's look at 2nd example of the effects of recombination.

Example 2: Siblings Matching Differently 
Sometimes a distant cousin will match two siblings at different cM levels and on different chromosomes.

Family Case 2: Match B
Michael Mitchell (sibling) & Match B
Shared cM -> 47.69
Longest Segment -> 25.55
Chromosome 3 -> 25.55cM
Chromosome 6 -> 22.14cM

Muriel Mitchell (sibling) & Match B
Shared cM -> 10
Longest Segment -> 10
Chromosome 3 -> 10cM

Steve Handy Jr (myself) & Match B
Shared cM -> 0cM
Longest Segment -> 0cM
No ancestral cMs on any of my chromosomes

     Shown above is an example of a Match B on my maternal side. Muriel Mitchell is my mother.  Michael Mitchell is my maternal uncle. Match B shares 47cMs with my maternal uncle and 10cMs with my mother. Both Michael and Muriel Mitchell are siblings. The key is the last recombinational event at the sibling level. The last recombinational event occurred separately between my maternal grandmother and each of her children. Genetic recombination randomly removed a large chunk of ancestral cMs in my mother's line. From maternal grandmother to the ancestor, the line of descent is one and the same. However in the last generation from my maternal grandmother to my mother, a single recombinational event removed a large chunk of ancestral cMs.

     On the other hand, from my maternal grandmother to my maternal uncle (mom's brother), 47.69cMs remained which was enough for the matching algorithm to detect it.

    FTDNA's matching algorithm doesn't report anything less than 20cM. Therefore the Match B actually didn't show in my mother's FFinder match list. I had to look at the FTDNA chromosome browser to see the amount of overlap which was 10cM. As one can see, siblings can match a distant cousin at different cM levels. By the time I arrived on the scene, recombination completely erased all traces of the ancestor's matching DNA from my chromosomes.

    One of the biggest misconceptions out there is that an autosomal DNA test can work against a group of people due to their ethnic background or history. As one can see, this is simply not the case and is actually quite impossible to do. 

There are two important biological events to remember in this tutorial.

  1. Genetic recombination indiscriminately and unbiasedly "chops" up DNA in each successive generation as a child is being conceived. 
  2. Each parent then randomly passes on 50% of DNA to a child while the parent retains the other 50% of their own DNA
    The two biological processes that were just mentioned are simply too impossible to control and to predict.

     The only time a population's ethnic background, history, or any other social factors is relevant in an autosomal DNA test, is if there is a history of inbreeding among close relatives. For example, people who are of Ashkenazi Jewish ancestry are descendants of ancestors where many first cousins married each other and produced children. This also occurred in early colonial America as well. This was, and still is, a social norm in many cultures. In examples of this nature, an autosomal DNA test will show distant cousins that share an unexpectant large cM amount of DNA. The reason for this is that the population's gene pool has became highly homogeneous under such inbreeding conditions over time.  Recombination will simply dice up more similar DNA segments and more offspring will receive those similar DNA segments.

     Even in light of such historical inbreeding, genetic recombination operates in the same fashion as it always does. DNA is randomly chopped and dispersed to descendants in an unbiased, random, and uncontrollable fashion. In the end, the matching algorithm is left to deal with the "scraps" and make a decision. The autosomal DNA matching algorithm always operates within an area of uncertainty. It merely looks at the resultant DNA scraps and leftovers from recombination, declares if there is evidence of common ancestry between parties, and gives a prediction of the actual relationship between two or more parties.

Well that's it for recombination. Hopefully that will clear up any misconceptions.

As always, it was a pleasure to serve everyone.



  1. This is one of the best articles I have read on atDNA ----Thank you for the share !!!

  2. Thanks for sharing...It helps to understand recombination

  3. Very clear and informative presentation. Thank you very much.
    Christer Amnéus

  4. My dad is 3/4 irish. My mother has some irish. Her mother was mostly Scottish. Her father was from the South. There was much inbreeding. In fact. several families inbred over the generstions crossing from each others family lines inbreeding esch back into their own family. Very confusing, indeed. My question is why on a dna test the results show thst I am only 20% irish scott mix, but. 24% scottish. The rest german, swedish, and french. Should I not be a straight, at least, quarter irish, or did inbreeding on my mother's side some how affect. he results on my father's side?

  5. This is an excellent explanation. There is a genealogist who has been working with me, and I have been questioning this formula he uses to determine cousinship. This confirms to me, that his formula may not work for everyone he works with. I have recently had a breakthrough, and believe I have finally discovered my specific family branch, and two of the cousins I have already discovered are estimated by FTDNA to be 2nd-4th cousins, total cMs 80 to 88, and I have a long segment of 43cMs with the 80cM match, but I have more short segments and a few mediocre segments with the 88cM match, and they are 1st cousins, because their fathers were brothers. They decend from their gggrandfather, and I believe I descend from his brother, which is my ggggrandfather, so our relation with actually be 4th cousin, once removed. Thanks again for confirming some things for me, Doug

  6. Thanks, this actually helped with my homework.
    Have a great day!

  7. I read five articles and watched six films today on recombination. This is the one that made sense. Thanks so much for writing this and helping me to understand.

    My only negative comment is to change the word "inbreeding" for something else. "Intermarrying" perhaps. Because it has a very negative connotation.

    1. Thank you. Point taken. I will remove the word interbreeding and replace with intermarrying

  8. Very informative, thank you. Could recombination explain why, when I examine my raw dna data, I've found instances where some people who are not apparently related to each other (I've done in depth checks) match me at the same locations on a chromosome. I have matches on my maternal side that come from known people from both her maternal and paternal lines, and also some that come from known people on my paternal lines. This is not limited to any one chromosome or location. Or is it more to do with 2 strands of the same chromosome from each parent? help! Jenny

  9. MGM Grand Casino Hotel and Spa - MapYRO
    MGM Grand 경상남도 출장샵 Casino Hotel 이천 출장안마 and Spa is located on the north side of the Las Vegas Strip, across from 천안 출장안마 the Las Vegas 성남 출장샵 Convention Center and features 창원 출장마사지 a casino,