Grandparents as reality, not theory

By Razib Khan | April 17, 2012 9:18 am

I am not particularly mystical or sentimental about genetics. I favor openness. But I just started getting my daughter’s results back from 23andMe, and some of her coefficients of relatedness to her grandparents deviated sharply from 0.25. As I have blogged about this possibility I was obviously aware of the abstract probability here; but it is a different thing altogether to be faced with reality. How exactly does one go about explaining that one of your parents is ~50% closer genetically to their grandchild than the other? I don’t think it matters really in a concrete sense for them, but divulging this information makes me somewhat uncomfortable. Many, many, others are going to be confronted with these issues. We don’t have social norms yet. This isn’t cut & dried like paternity. Thoughts?

  • DNA-Aleu

    Thats indeed a good question. Better not to make testing on children to avoid this uncomfortable situations? But humans want to know and understand, thats how science is going on. And also evolution needs facts like this one.

  • Razib Khan

    #1, well, that seems a trivial reason to NOT get tested IMO. but in any case, the point is that honestly am taken aback by how to explain the results to my parents (who lack a genetics background) without sound weird or insulting.

  • Stephan Guyenet

    I remember in The Selfish Gene, Dawkins talking about genes that had evolved ways to favor their own inheritance in a non-Mendelian fashion. Seems like the selective pressure would be very high to evolve ways to circumvent normal Mendelian inheritance. There is obviously some balancing selective pressure that prevents this from becoming widespread, but it does crop up from time to time evolutionarily speaking. I have no idea what the molecular mechanism could be though.

  • Violet

    Perhaps you should promise to the lesser contributing grandparent that you will try to have more offspring until the average gap evens out? :)
    This is only the first flip of the coin?

    Also, in my family the lesser contributing grandparents are often more fond of the grandchild because they see all the characteristics that made them to love their spouse.

  • Razib Khan


    this is genome-wide though. so it is normal segregation and recombination.

  • Razib Khan

    #4, well, luck that my parents are not divorced. but then, you know much big of a factor love is in most traditional south asian marriages…. perhaps i will introduce her to the less related side of the family? :-)

  • EWCZ

    Wasn’t that so even in pre-genotyping times?
    Especially when young, I looked very similar to my paternal grandmother and my relatives commented on that. True, there was no exact measure of relatedness a t that time except the phenotype, but the general concept of a grandchild resembling one particular grandparent is not new.
    I wonder, if the relatedness coefficients of your daughter will be matched by her later appearance as well.

  • toto

    #4 beat me to it. Only solution is to increase N. :)

    /Younger brother got three kids in about four years.

  • miko

    I’d be fascinated if grandparents can infer different degrees of relatedness among their grandkids and bestow resources accordingly. I’m just saying… my paternal grandparents have 12 grandchildren and we are not all equal in their eyes, and it seems to be from a very early age. It also seems to correlate with physical similarity.

    Other monkeys seem to be able to tell

  • Junior

    Exactly as EWCZ said. We already resemble more to a parent and a grandparent than the other. Knowing this to the genetic level shouldn’t make you feel any different. I strongly disagree with those who would rather not know or not to release the information to protect one’s “sense of self”. What “miko” is proposing I do find it unethical though. But yes, grandparents should be entitled to do as they please with their own resources, even if they are making a stupid decision.

  • Razib Khan

    #10, i think it is kind of hard to see resemblances at such a fine grain at infant age. but perhaps you have a keener eye than i.

  • pconroy

    Are you getting this number from the 23andMe Inheritance Calculator, or from independently looking at her raw data?

    My eldest daughter is listed as 29% paternal grandfather and 21% maternal grandmother – as opposed to 25% each – so my father is responsible for almost 40% more of her inheritance, through me.

    Of course, you need to realize, this is of the total number of SNPs tested by 23andMe, not of her total genome – which may be nearer to 25% each!

    I shared this fact with my daughter – but didn’t bother telling my parents, as they wouldn’t understand at all.

  • Razib Khan

    @12, yeah. but the issue is that this is 1 million markers over the autosome. i doubt it’s THAT unrepresentative.

  • Siod Beorn

    What are the valid assumptions that you can make if one grandparent is responsible for more of your child’s genetic inheritance?

    I think the uncomfortable assumption is that the child is less theirs–like, but not, an adopted child. However couldn’t the lower genetic inheritance be as, or more, important than the other grandparent’s contribution? I.e., it packs more punch in the areas that really matter?

    Either way, you could emphasize the fact that it’s still a significant contribution; that the child is somewhat representative of them.

  • Razib Khan

    so one thing here: don’t over-rationalize this. my parents don’t know what coefficients of relatedness mean, and don’t care. rather, the issue is how do i communicate this colloquially without sound weird? i can tell ’em the relatedness values and they wouldn’t care. and i can present it in a noninformative anodyne colloquial manner. but if i’m going to tell someone colloquially it just sounds rude and weird in my head.

  • Charlotte

    Razib, just say the 23andMe results show she “takes after” one of the grandparents.

    It’s something they were expecting to hear anyway. I “take after” my father, who “takes after” his Uncle John — the facial resemblances are striking in photographs.

    If you want a stronger term, you can say she’s the “spitting image” of grandparent X. One of my sisters is the “spitting image” of our great-great-grandmother (mother’s mother’s mother’s mother).

    All families know about these things.

  • Siod Beorn

    Oh, then I was reading more into post than you put into it–thought you were more concerned with how they would react. I second Charlotte’s suggestion, then.

  • pconroy

    Well with 4 grandparents sequenced, you must have a 1 to 4 ranking of contributions per grandparent. So I can’t see a problem with simply saying:

    ” X ought to inherit 25% of her ancestry from each of you 4, but actually got a tad more from grandparent A, and a tad less from grandparent B, and overall her ancestry is A %, B%, C%, D% from you guys”

    “Tad” being an ambiguous measure?!

  • marcel

    @pconroy (18): But traditionally “tad” has referred to boys and Razib has a girl ,so we have to come up with a different ambiguous measure, alass. I think “a petite bit more” fits the bill, equally ambiguous, but clearly feminine. More seriously, I endorse Charlotte’s (@16) idea of “takes after”

  • Dm

    What to say to the grandparents now, it should strongly depend on what you said when you asked them to submit their DNA samples?

    If they were hooked on some ancestry myths or a quest for lost family branches, then I assume that they didn’t gain much from one grandchild’s result … better phasing perhaps? If they worried about nonpaternity, then they could be reassured. If they were curious which traits they may have passed to the baby (as I would have been in their place) then you need to investigate that … and the overall percentage doesn’t matter. And if they had no interest at all but you badgered them and paid for the tests, then just tell them that so far they were 100% right, that the money’s been busted for no apparent benefit, that of course the baby shares a lot of genes with each of them, as everybody would have guessed in the first place … but that, as the time goes by and as the mankind’s knowledge grows, their grandchild will be able to extract many useful bits of knowledge from these results … so the investment will pay off, in time. And they will chuckle, yeah right.

  • Nandalal Rasiah

    That’s like the awkward question I’ve received while looking at baby pictures of friends, “do I look more like mom or dad here?” It’s so common in the US, I’m not sure it will be awkward at all to explain that more material comes from one grandparent as opposed to another. also, given the nature of south asian arranged marriages, they might not be emotionally opposed to the idea that processes beyond their now- crystallized intelligence have real-world impact and has no reflection on their individual worth.

  • Charles Nydorf

    Whatever you tell people, they’ll make up their own minds based on her phenotype or their impressions of it. One my father’s side I had one uncle who always worried that I did not look anyone in “the” family while others were struck with how much I looked like my paternal grandfather. On my mother’s side, relatives were almost desperately eager to seize on some family similarity, in voice or what not. The funny thing is that the relatives whom I was really close to never talked about any of this.

  • Doug1

    Razib, just out of curiosity, is your daughter’s genetic relatedness about equal to each of her maternal grandparents – your wife’s parents?

  • AG

    Similar to the outcome of offspring’s gender, some time we really do not have much of choices. Maybe gender is the best analogy

  • Brian Too

    Charlotte’s suggestion is the way to go IMO. It gets the idea across without raising needlessly troubling issues and questions. It opens the door and if they want more information they can ask.

    It also has the great merit of correlating phenotype and genotype in a familiar, comfortable way. I know that if I used words like “phenotype” with my grandparents, I would immediately get hung up on the rocks of grammar, semantics, educational background and all the rest. Plus they would probably say I was getting too big for my britches!

  • Razib Khan


    paternal grandfather = 0.29
    paternal grandmother = 0.21
    maternal grandfather = 0.23
    maternal grandmother = 0.27

  • Stephen

    Lets get down to possible actual reasons for the data. I don’t know enough to puzzle it all out, but here are three things to consider. 1) What is the company’s kinship algorithm? How does it handle common versus rare markers? 2) There aren’t a million independent units of inheritance being sorted beanbag-wise here. Big blocks of SNPs are inherited together, making the assortment much more stochastic than random individual sampling. 3) I understand that there are diverse origins of the ancestors involved here. So some grandparents may share haplo-blocks that are rare on a global scale. SO, all of these things might well interact to reasonably produce the actual data. Can’t solve it here, but just point to the space. Are the data even surprising? Compared to what expectation, based on what pre-suppositions, especially about effects of recombination on the sampling?

  • Razib Khan

    #27, it’s just segregation (stochastic sampling). i’ve blogged this before. not only that, some of the chromosomes are far bigger blocks than others. recombination works against this.

  • miko

    It can’t be that they are just giving the proportion of shared SNPs, because then they would add up to more than 1. They must be inferring haplotypes and identifying recombination events. The proportions are chromosomal contribution. I always assumed for relatedness they just report shared SNP %, but here they must be using pedigree as well. Or did I miss something?

  • Stephen

    #29 – The proportion of shared SNPs could be normalized somehow. But if it really is the proportion of fully mapped chromatin length from each ancestor, that’s pretty cool. It’s much bigger computationally, but that’s their business. And if so, the fact that it’s not 25% across the board is not so much surprising or begging explanation. It’s just very instructive.

  • Razib Khan

    #29, yeah, pedigree. there is a shared SNP feature that they are getting from plink, but that’s now what i’m talking about here.

  • Peter Ellis

    I agree with Charlotte (#16) – but also note that unless they ask/want the precise numbers, I don’t think you have to give them.

    Best option would be to back it up with some functional information: she “takes after” one or other grandparent in some of her blood group / taste receptor / stature genes, etc. Suitable functions to talk about are ones that are non-visible – they already know her hair colour, after all – but with no real medical impact. That way each grandparent gets to feel kind of cool about what phenotypes they contributed to their progeny.

  • stuart

    Charlotte is right, and this IMO is a non problem. Children general have a physical resemblance to a parent or grandparent. My girlfriends son looks like his father who in turn looks like his father and I would bet anything his markers would match closer to both of them, and the resemblance is not just physical but personality traits are there too. However her 2 daughters resemblance of family members on their mothers side is uncanny. I think three results are to be expected.

  • Mikey Mike

    I have only one grandparent in my dataset on 23andMe so far, and he’s ranking 29.5% on a grandson, and 26.1% on a granddaughter. He’s the maternal grandfather.

    Inheritance is like pulling cards from a deck, with each of your parents having cards from their parents (the grandparents generation). Females will get half of each parent’s genes, but you don’t know which half (which cards), which explains the imbalances. Males will skew a bit more than half to their mom’s because the X chromosome is a lot bigger than the Y.

    FWIW, the grandson (he’s an adult) is nothing like his maternal grandfather.

  • Razib Khan

    #33, please remember that there is no variance in inheritance from the parents. on the autosomes parents give you 50%, each, exactly. so resemblance to either parent has to do with specific issues of genetic architecture (e.g., dominance).

  • Eurologist

    It’s like throwing dice – so the only way I can think of communicating this is play and humor. And, since the deviation from 25% is not all that large (to a lay person), I doubt grandparents would mind. Each are still represented by more than 1/5. Start with saying how approximately equal the shares are!

    paternal grandfather = 0.29
    paternal grandmother = 0.21
    maternal grandfather = 0.23
    maternal grandmother = 0.27

  • Dm

    Eurologist, God isn’t really expected to play dice. My limited understanding of science-religion interplay in Islam makes me believe that the laws of Nature work though God’s will to enforce the behavior of the material objects in accordance to His pre-defined rules. But if a law is understood by the scientists to be a game of chance rather than a mechanistically predetermined act, then what would the adepts think of the divine will? It may be easier to understand the inheritance in a different way, as a series of specific pre-ordained events. The crossingover happened here and there and there; the chromatids went this way in meiosis; and these gametes went on to start a new life. This way of explanation would require no chances and no surprises about the actualities being somewhat unequal or perhaps unfair. It wasn’t a mere chance, you’d say. It is the reality, and therefore, a reflection of God’s will, and of our humility.

  • Razib Khan

    My limited understanding of science-religion interplay in Islam makes me believe that the laws of Nature work though God’s will to enforce the behavior of the material objects in accordance to His pre-defined rules

    more dominant in sunni islam, fwiw (i.e., they position re: god’s will approximates calvinism). though not universal, and far less so among shia.

  • Christopher@BorderWars

    You can also point out to your mother, assuming that contributing fewer SNPs than your father is an issue, that SHE is the reason that her granddaughter is a female given that it’s her X chromosome that you passed on to your daughter to determine her gender.

    That’s a fair trade, no?

  • pconroy


    It’s interesting that you have:
    paternal grandfather = 0.29
    paternal grandmother = 0.21

    As that’s what I have with my eldest daughter.

    Do you think that this represents some sort of paternal imprinting mechanism, where some chunks of DNA are preserved intact, or this is totally random? I’m thinking of IGF-2 here:


Discover's Newsletter

Sign up to get the latest science news delivered weekly right to your inbox!

Gene Expression

This blog is about evolution, genetics, genomics and their interstices. Please beware that comments are aggressively moderated. Uncivil or churlish comments will likely get you banned immediately, so make any contribution count!

About Razib Khan

I have degrees in biology and biochemistry, a passion for genetics, history, and philosophy, and shrimp is my favorite food. In relation to nationality I'm a American Northwesterner, in politics I'm a reactionary, and as for religion I have none (I'm an atheist). If you want to know more, see the links at


See More


RSS Razib’s Pinboard

Edifying books

Collapse bottom bar