The story behind this post is that my wife recently gave birth to our first son and we experienced a funny story about genetics the day following the birth. Before I start I should say, to reassure the reader, that I have no doubt that I am indeed the father of my child. But as you will see, a non-geneticist might have become worried when faced with the same situation.
Firstly, my wife has a negative rhesus type. This has important medical implications because if the baby were to have a positive rhesus type, she would create antibodies against this marker which could be life-threatening for any subsequent child of positive rhesus type. Basically this is a relatively big deal, but there are ways to deal with this, and therefore knowing the blood type of the baby is essential.
The day after the birth, while we are both lying on our bed, very tired, a midwife comes by and asks us whether we know the rhesus status of the baby. We answer negatively, she checks her notes and says, “Ah, good news, the baby is rhesus negative. The father must also be rhesus negative then!” Well, I am not…
At this very moment, as confident as one can ever be that there is no paternity issue, one cannot help wondering, even for a fraction of a second. The answer is: “Don’t panic”. But still, some harmless checks cannot hurt, right?
What do I know about rhesus type? Not much. The rhesus blood type is defined by a deletion in the RHD gene. Rhesus negative is a recessive trait, which means that my wife’s genotype is -/- (in other words, both of her copies of the RHD gene are deleted). On the other hand, as a known rhesus positive individual I could be +/- (one active and one deleted copy of the gene) or +/+ (both copies of the gene being active). It turns out that I have studied this deletion somewhat in the context of my involvement with the Wellcome Trust Case Control Consortium (WTCCC). In fact we thought for a while that the negative rhesus type was associated with rheumatoid arthritis (until the data proved us wrong and we eventually gave up on that idea).
Now, of interest to this blog, we also looked in this study at single nucleotide polymorphism (SNP) markers that tag this rhesus deletion (the result of this work with the WTCCC can be obtained here). SNPs are exactly the type of markers that 23andMe looks at and therefore some information on rhesus blood type is available in my 23andme data. Our code name for the RHD deletion is CNVR116.1 and the WTCCC file shows that the best tag SNP we have based on our genome map (the HapMap) is rs873308. This SNP rs873308 is indeed a decent marker for the rhesus deletion. In our technical jargon we say that the r-squared is 0.55 (with the RHD deletion causing the negative blood type) and this r-squared value, essentially a correlation coefficient between both markers, is scaled between 0 and 1 (1 being the best possible tag). Note that the lack of an ideal SNP tag (r-quared > 0.9) was an issue for our work in the WTCCC experiment and it cost us some time to assess the rhesus type of our collection of cases and controls.
To obtain the most accurate understanding of the correlation between rs873308 and the RHD deletion I would like to have a large panel of individuals with genotype data for RHD and rs873308. Unfortunately, I don’t have it readily available (even though this probably can be found somewhere). I do have, however, genotype data for another good but slightly less efficient tag rs10903129, which is also on the 23andMe SNP panel. The reason why I have these data is because the same individuals typed for the RHD deletion in the the WTCCC CNV experiment were also typed for SNPs in this related (and initial) study, and this rs10903129 was part of this SNP genotyping array (the Affymetrix 500K). For rs10903129 the r-squared value with the RHD deletion is 0.49. Not ideal, but something relatively good.
Now what do the data tell me? My rs10903129 genotype is AG, which means that I am a heterozygous individual (as opposed to AA or GG, which are homozygous genotypes). Combining existing data for RHD and rs10903129, I find that individuals of European ancestry with the AG genotype (like me) have a 70% chance of being a +/- rhesus individual (as opposed to being +/+ or -/-). In fact, because I know I am not a -/- individual (because my rhesus type is positive) I have a slightly higher probability of being a +/- individual (71% as opposed to a 29% probability of being +/+).
What does this mean for my simple paternity test? As a 71% chance +/- individual, I may very well have passed the negative (-) rhesus type to my son, who also received a negative rhesus type from his -/- mother. Therefore, I have no reason to doubt that I am indeed the father of my son (not that I had any reason to start with!). But if a reader, looking at my data, can convince me that I am indeed a +/+ rhesus type then well… a personal email rather than a public comment would be appreciated of course!
The rather obvious conclusion from this practical genetics experience is that 23andMe genotype data can be used as a paternity test. And of course there are much better ways to do it: the rhesus type represents only a tiny fraction of the available information, but a particularly obvious one. It also seems to me that if genome sequencing at birth becomes a routine test in the future, it probably means that we will have a systematic paternity test for each newborn. I find it to be a rather significant change, and whether it is a good thing is definitely arguable.
And most importantly, midwives making random comments about genetics should know better and not make any comment that is not required. I can imagine that a statement of this type could make a father quite suspicious for no good reason. This being said if both parents have a negative rhesus type (-/-) and the child has a positive rhesus type (so probably +/-) then… well, it’s hard to see how this could happen if the father is indeed the father. But even in this case, I suppose that even a very standard and highly reliable test like rhesus blood type can be mistaken in rare occasions so even this situation is not completely conclusive.