8

If a parent (Charlie, a male) has been autosomal tested, and does not have either of his deceased parents (Alice and Bob) tested, and thus he does not have his autosomal DNA phased between their mother (Alice) and father (Bob), does testing one or more children of Charlie (Dick, Elizabeth, Frank, and George) help with phasing Charlie?

What if some of Charlie's children (Frank and George) are with a different spouse of Charlie than the others (Dick and Elizabeth), does that help more?

Are there any tools to help with this phasing like we see with GEDmatch giving phased results for Charlie when a parent (Alice or Bob) is tested only we want to phase Charlie, using the aid of his tested children, Dick, Elizabeth, Frank, and George.

  • 1
    Please clarify who is who's children. Regarding your questions, you can't phase if no parents kit are available. It is non-sense. You can try to recover unavailable kit based on close-relatives kit. There is such great tool on gedmatch called lazarus. – George Gaál Aug 5 '17 at 22:12
  • @GeorgeGaál I've tried to clarify the question, let me know if that is still not clear. A Lazarus kit does not help in this case. – WilliamKF Aug 6 '17 at 0:53
6

Edit: Brainstorm and I had an extended discussion through the comments on whether Visual Phasing of siblings can add information that the parent's DNA test doesn't already add. I have been convinced that yes, doing so can provide information not had by the parent alone. It is possible to phase the parents from the children with the additional help of matches of other relatives that will identify the parent and grandparent that each segment of DNA belongs to.

So yes, Brainstorm is correct and there is a benefit to DNA testing children. If you have 3 or more children and test at least 3 of them, they can be used to Visually Phase most of both their parent's DNA, with or without the parent being tested. One or two won't be enough to Visually phase, but they can still provide some bits of help in other ways.

My apologies for giving Brainstorm a hard time. My thanks to Brainstorm for being persistent and letting me work this out and learn something important.

I am deleting my incorrect statements below, and keeping just what's left that is useful.


If the parent is tested, using the child's matches is of little help, since almost all valid matches that the child has from that parent's side, his parent will also have. With that said, there is one place where a child tested may give a small help with a parent's matches. In the cases where both the parent and the child match the same person, this double match provides an extra confirmation that the parent's match is real and is not a match by chance. This could help some of the time (only about half of the parent's matches will also match with a child) and only in 7 cM to 15 cM segment range, since single matches 15 cM segments are almost always real (Identical by Descent) without confirmation, and double matches below 7 cM still can be by chance.

Much better is to try to phase the parent into his maternal and paternal side. It can be done with relatives, as George Gaál said in his answer about phasing using the Lazarus tool at GEDmatch to help you phase your parent's DNA using relatives of his parents.

Or it can be done through Visual Phasing as Brainstorm answered. Doing this will make use of the children's DNA data to add information (help phase) the parent, and matching with a phased kit will provide better matches to help find the parent's DNA ancestors, so Brainstorm is correct.

Regarding your question if it would help if some of the parent's children are half-siblings with different mothers: if anything, that will make the phasing more difficult, because there will be three parents to determine, not just two. I've not seen any articles of anyone attempting to visually phase using half-siblings.

Phasing techniques work with the raw data. It is quite involved. But once you phase a parent into their paternal and maternal side, you can load the phased parent into GEDmatch, and find better matches that are only on the one grandparent's side that are less likely to be matches by chance. This will definitely help you find your parent's DNA ancestors.

The other thing you can do is to find a few cousins of the father who have tested so you can triangulate matching segments with other people. Triangulations that don't include matches by chance are often Identical by Descent and can help to identify a common ancestor for the three people. But triangulating with the children is of no help. Anyone the child and the cousin triangulates with, the parent and the cousin will also triangulate with.

If you don't want to do any heavy analysis, then if you tested your parent at FamilyTreeDNA (if not, you can transfer your raw data there) you can create a tree of your known relatives. Then you can link any relatives who have also tested their DNA at FamilyTreeDNA. This will partition your parent's matches into paternal, maternal or both, and you may find that useful in separating out your matches. Notice in the screenshot below, my 11,878 matches are segregated into 3,024 paternal, 100 maternal, and 19 both, and are designated by the blue, red or purple icons which are shown below the name of each match. If I had more relatives on my mother's side tested, then by linking them, the maternal count would increase. Note that testing a parent's children will not help here.

enter image description here

| improve this answer | |
  • I pseudo-phased my parents using the tools from David Pike. I then used my parents' pseudo-phased DNA to recreate 50% of each of my grandparents' DNA. Here is how. I used David Pike's tool to phase DNA for me, my brother, and my parents, and also identify recombination points for my brother and me. I created my evil twin (my parents' DNA that I did not inherit). I uploaded my DNA, my evil twin's DNA, and my parents' DNA to FTDNA. I looked at matches around the recombination point to identify whether it was from my DNA or my brother's DNA. Gma and Gpa are on opposite sides of the recombination. – Brainstorm Nov 30 '17 at 5:05
  • @ikessler - The cool thing about FTDNA is that it is full of relatives - even ones you didn't know you had. It is pretty easy to figure out at least some matches for each of the four grandparents. Try it - you will be surprised at how easy it is. For my dad, I was able to identify the DNA he inherited from his father and the DNA he inherited from his mother. I did the same for my mother. Why do you say that is not phasing my parents? I think we must be using different definitions of "phasing." – Brainstorm Nov 30 '17 at 5:50
  • @Ikessler - my bad. I thought the question was: "Are there any tools to help with this phasing like we see with GEDmatch giving phased results for Charlie when a parent (Alice or Bob) is tested only we want to phase Charlie, using the aid of his tested children, Dick, Elizabeth, Frank, and George." Phased data won't match more relatives than unphased data. But some people (including me) think that phasing is helpful to identify how relatives are connected. I believe it helps you sort out the relationships, which makes it easier to find common ancestors (including new ones). – Brainstorm Nov 30 '17 at 6:36
  • @Brainstorm - No! Me bad. You right. Q is: "does testing one or more children of Charlie (Dick, Elizabeth, Frank, and George) help with phasing Charlie?" – lkessler Nov 30 '17 at 8:06
9

I think the prior answers are either misleading or wrong, and people should not miss out on the opportunity to get great DNA information because they were discouraged by those answers.

If you have the DNA for Charlie (the parent), and any of Charlie's children (Dick, Elizabeth, Frank, and George), you can at least partially phase a child (and thus pseudo-phase the parent too). An example of a tool that does this is found here:

http://www.math.mun.ca/~dapike/FF23utils/trio-phase.php

And here is a link to a tool you should use when you have DNA for more than one child:

http://www.math.mun.ca/~dapike/FF23utils/phase-sibs-2parent.php

Ideally, you would use data from both parents to phase the DNA of the child. However, both of these tools can run with data from only one parent. The caveat is that you will probably end up with "no calls" at some of the SNP locations.

If you want to do the phasing manually, the Journal of Genetic Genealogy has an article titled "Phasing the Chromosomes of a Family Group When One Parent is Missing," available here:

http://www.jogg.info/pages/62/files/Athey.pdf

By the way, you can phase Charlie's DNA using only the children. Of course, any DNA that Charlie did not pass to his children would not be reconstructed. Here is a link that explains how to phase a parent's DNA with only the DNA from the children:

https://thegeneticgenealogist.com/2016/11/21/visual-phasing-an-example-part-1-of-5/

These links explain the concepts much better than I can in this short answer. But here are the basic principles: By phasing the child, you will know which DNA the child inherited from Charlie. You will also know which DNA Charlie did not pass on to the child. At this point, you have already pseudo-phased Charlies DNA. Now you just need to know the recombination points for the DNA that Charlie passed on to his children. You can find the recombination points by comparing the children's DNA with each other (see the visual phasing article mentioned above).

Testing the children does not help identify new DNA connections, but it does help to determine the relationship of the existing DNA connections. First, testing can be used to phase Charlie's data, which then tells you which of Charlie's parents share the DNA with the DNA connection. Second, the phased DNA gets rid of many false matches (the IBS matches) and the remaining matches are more likely to be valid (IBD).

Hopefully you can get Charlie, Dick, Elizabeth, Frank, and George tested before it is too late. I would have them all tested, because you can probably reconstruct and phase most of the DNA of their mother(s) as well.

The original question also asked "What if some of Charlie's children (Frank and George) are with a different spouse of Charlie than the others (Dick and Elizabeth), does that help more?" The answer is yes, especially if that means you are adding a second or third sibling. There are also benefits to adding a fourth sibling, but they are not nearly as great as adding the second sibling (allowing you to identify recombination points) or adding a third sibling (allowing you to identify which sibling is responsible for each recombination point). The fourth sibling may help resolve some no-calls in the phasing of the other siblings or provide additional DNA to reconstruct a missing parent.

Edit: I was asked to provide an explanation of an actual example of phasing a parent using DNA from the parent's children.

Here is the procedure I used to completely phase my parent’s DNA. This procedure also extracted one of the two sets of 23 chromosomes (albeit pseudo-phased, not fully phased) for each of my grandparents. That means I identified 50% of the DNA (approximately 3600 centimorgans) for each grandparent.

Step 1: Mail spit samples from me, my brother, my father, and my mother to Ancestry.

Step 2: Download Ancestry DNA raw data for each of the spit submissions. It looks like Ancestry recently changed the format of how it provides raw DNA data. I obtained raw DNA data last week. As of November 29, 2017, FamilyTreeDNA would not accept my raw data in the format provided by Ancestry. Hopefully FamilyTreeDNA figures this out and provides a solution soon. My parents obtained their DNA from Ancestry a few months ago, and FamilyTreeDNA accepted their data without any problem. I made my data file look like my parents’ data file (explained in Step 3 below), and it uploaded fine. My parents’ files are 668,962 lines long, while my file was only 650,430 lines long. It looks like Ancestry is now providing results for a slightly different set of SNPs.

Step 3: Create a new raw data file for me and my evil twin (my parents' DNA that I did not inherit). I wrote a Perl script that read in my father’s data and created an index for each RSID. The script then read in my mother’s data and added her data for each RSID. The script then read in my data. If an RSID in my data was also an RSID in my parent’s data, the script printed the alleles I inherited from parent’s to my file, and the alleles I did not inherit to my evil twin’s file. If an RSID was in my data but not my parents’ data, it was ignored. If an RSID was in my parents’ data but not my data, the script printed a “no call” (“0” instead of G, A, T, or C) for each allele in my file and my evil twin’s file.

Step 4: Upload the new raw data file for me, my evil twin, and my parents to FTDNA.

Step 5: Pay the $19 for each kit to get access to the chromosome browser. For each kit, go to the chromosome browser page and click on the link to “Download All Matches to Excel (CSV Format).”

Step 6: Copy the downloaded data from all four .csv files into a single spreadsheet. Sort the data by 1) chromosome; 2) start location; 3) centimorgans

Step 7: Use David Pike’s tool to identify the recombination points in the DNA for me and my brother.

http://www.math.mun.ca/~dapike/FF23utils/phase-sibs-2parent.php

If you have three siblings tested, it would greatly simplify the identification of which sibling was responsible for the recombination point. If one sibling’s pattern changes while the other two siblings’ patterns are the same, the one that changed must be responsible for the recombination point. With only two siblings, you have to rely on DNA matches (explained further below) from the chromosome browser data that you downloaded. I had to do it the hard way with two siblings.

Please note that Ancestry provides its data in Build 37.1 format. Accordingly, the output data from David Pike’s tool will also be in Build 37.1 format. But the FamilyTreeDNA chromosome browser data is in Build 36 format. It is important to convert the recombination point from Build 37.1 to Build 36.

Here is a link to a tool that does the conversion:

https://www.ncbi.nlm.nih.gov/genome/tools/remap#tab=asm&src_org=Homo%20sapiens&src_asm=GCF_000001405.13&tgt_asm=GCF_000001405.12&min_ratio=0.5&max_ratio=2.0&allow_locations=true&merge_fragments=true&in_fmt=guess&out_fmt=guess&genome_workbench=true

David Pike’s tool shows that either my brother or I had a recombination point on Chromosome 1 of my father’s DNA at rs6426802,rs28497538 (locations 19202507-19205672).

I entered Chr1:19202507-19205672 into the “Data” box of the conversion tool (see link above), and the conversion tool showed that the recombination was at locations 19075094-19078259 in Build 36.

I filtered the matches in the spreadsheet to show matches on chromosome 1 with a start location less than 20000000 and an end location greater than 18000000. My dad had two fairly large matches that spanned the recombination point. The first match was from 8109798-38365641 (45.55 centimorgans). That same individual matched my evil twin from 9316595-19061043 (19.52 centimorgans) and me from 19063913-38365641 (24.1 centimorgans). The recombination point is mine, and not my brother’s. It is interesting that the match start and end points don’t align with the recombination locations at 19075094-19078259. I haven’t figured out why there is a mismatch in the locations. Looking at the raw data, I think David Pike’s tool got the recombination point correct.

David Pike’s tool shows that either my brother or I had a recombination point on Chromosome 1 of my father’s DNA at 55495744-55518467. This corresponds to 54853453-54876176 in Build 36. I filtered the matches in the spreadsheet to show matches on chromosome 1 with a start location less than 60000000 and an end location greater than 50000000. My dad has a match from 38368451-62971860 (26 centimorgans). I match that individual from 38631015-62971860 (25.45 centimorgans). This means that my brother has the recombination point at that location.

Sometimes there will not be a match that spans the recombination point. You can still solve who is responsible for the recombination point by looking at matches above and below the recombination point. If I have a match on my grandma’s side above the recombination and my evil twin has a match on my grandma’s side below the recombination, the recombination point is mine. If I have matches on my grandma’s side both above and below the recombination point, then the recombination point is my brother’s.

I used the “Matches in Common” feature on FTDNA to identify the correct grandparent with the matching DNA segments.

Sometimes the recombination point cannot be refined to a single SNP. When reconstructing my parents' (and grandparents') DNA, I had to insert “no calls” for the range of possible recombination points. These ranges are usually small, and they do not impact matches if “no calls” are inserted in the correct locations.

| improve this answer | |
  • Sorry @Ikessler, but I don't think you have ever tried the process I described above. I have used it and it works. I don't understand the logic in your comment. Why does the fact that a parent gives different parts of DNA to a child mean that you can't use that information to phase the parent's DNA? Also, please note that the process does not fully phase the grandparents' DNA. Instead, the process results in pseudo-phased DNA for the grandparents. However, the grandparents' pseudo-phased DNA is identical to Charlie's fully-phased DNA. Try it next time before down-voting. – Brainstorm Nov 30 '17 at 4:04
  • You are correct. I humbly apologise. Yes, given children's DNA and Visual Phasing techniques, you can identify crossovers and thus the segments that match between children will be parts of parents and grandparents DNA that can be reassembled. Please make a tiny edit to your question so I will be allowed to take off the -1 and give you +1. I will update my answer to state that you are correct. – lkessler Nov 30 '17 at 6:40
6

I will try to explain more comprehensively.

What does phasing do and what task does it solve?

We use phasing to separate the list of matches of one's person into two smaller lists of probable relatives: by paternal and maternal lines. Also phasing is great in terms of eliminating IBS (non-IBD) matches. Imagine that somebody has small DNA overlap with your mother and small DNA overlap with your father. There are chances that your overlap with this person is greater than your parents have.

To be able to do phasing we need to know how to separate one's paternal and maternal lines. The best option is to get DNA from parent of person. The one parent is enough.

OK. You said that parents of Charlie are deceased and there are no option to get DNA from them. Then you need to other close relatives of Charlie. The brothers and sisters of Charlie won't help, because they share the same parents. Half-brother and half-sister won't help too. But the good bet is to get DNA from Charlie's cousin. Cousin shares with Charlie ONLY ONE parent. So there are big chances that they both have overlapped lists of relatives. More cousins - more precision.

Another good bet is the sisters or brothers of Charlie's parents. They share a great amount of DNA. But I understand that usually if parents are old, their sister and brothers are old too, so it is good fortune to get DNA from them.

I want to emphasize that the Charlie's children DNA won't help in phasing.

Also please read carefully about Lazarus tool on GEDMATCH. There is small guide how to use it: http://www.thegeneticgenealogist.com/2014/12/09/recreating-grandmothers-genome-part-1/

This tool uses the conceptions I explained before.

| improve this answer | |
  • 2
    I agree completely that testing Charlie's children's DNA won't help find Charlie's ancestors at all. However, testing Charlie's aunts, uncles, siblings or cousins might help in two ways: (1) Their DNA might, by chance, be a better match to other descendants of those ancestors, and, (2), aunt, uncle, or cousin tests would help sort Chariles' DNA matches into paternal and maternal groups. Aunts and Uncles would be best and are a pretty good substitute for unavailable parents. All of that is true before you even consider phasing. – Jamie Cox Aug 6 '17 at 17:43
  • @JamieCox, I totally agree with you :-) – George Gaál Aug 6 '17 at 17:57
  • I believe IBS chance can be reduced by comparing Charlie's matches to his children: Dick, Elizabeth, Frank, and George. If a match of Charlie named X matches Charlie and also Elizabeth and George match the exact same segment then doesn't that reduce the probability of IBS since the children only have have the DNA at any position and thus won't have a match that straddles back and forth across both DNA values at each position? – WilliamKF Aug 6 '17 at 18:25
  • @WilliamKF, what you say doesn't work. It is long answer why. Just trust me :-) For example, please refer to my GEDMATCH KIT T506702. One of my best matches is A684069. We share as much as 35cM. Very good, really? But my mother has only 8.5cM shared with this person. So I dag deeper and phased my kit resulting two kits: T506702P1 and T506702M1. And surprise! There is no shared DNA between T506702P1 and A684069 kits!!! So 26.5cM is artifact of test without NO REAL relation!!! – George Gaál Aug 6 '17 at 19:29
  • So. What is the practical result of such investigation? Now you know that it is BEST to test parents and their direct relatives, nor children. – George Gaál Aug 6 '17 at 19:30

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.