Real World Genetics


Chapter one of your genetics textbook will probably define "dominant" and "recessive", and that’s fine.  But it's not enough to get a handle on what’s really going on genetically with our dogs (or ourselves, for that matter).  Everything in the real world will make more sense if we take a moment to look at the genes themselves and the complications that arise to alter our simple Mendelian expectations.

First, what is a gene?

A gene is a length of DNA that codes for a particular polypeptide.  A polypeptide is a piece of a protein.  Polypeptides, and thus proteins, are made up of small molecules called amino acids.  Put the right set of amino acids together the right way and you get a protein.  Put them together the wrong way and you get, frequently, a nonfunctional or only partially functional protein.  For example, you might either have an allele that codes for functional tyrosinase, an enzyme necessary to make melanin; or else you might have an allele (the albino allele) that codes for a nonfunctional tyrosinase protein.  Failing to produce functional tyrosinase leads to a failure to produce melanin later down a particular metabolic pathway, and poof!  You are albino.  The important point is this:

Genes code for proteins, not directly for traits.

This matters because complicated traits, not to mention plenty that look simple, are probably influenced by a lot of different genes (they are polygenic; the genes involved are sometimes called polygenes).  Even “simple” traits may not turn out to be all that simple once you really look at them.  The common human genetic disease cystic fibrosis is caused by a failure of protein “channels” (cystic fibrosis transmembrane conductance regulators) in the membranes of certain cells to function properly in the transport of chlorine ions, which in turn means that the movement of water across those membranes does not happen properly.  Although this disease is thought of as a simple recessive, and is presented that way in every general biology text on the market, it’s not so simple in the real world.  Understanding the molecular biology behind gene action lets you see genes as they really are, rather than as magic wands that -- poof! -- create random effects.  This lets you make reasonable guesses about the genetic action behind specific traits you observe in your own dogs.

How does cystic fibrosis work in the real world?  The CFTR gene responsible for making the particular channel protein described above is subject to over a thousand described mutations (!) (, many of which can cause clinical cystic fibrosis.  The most common mutation that leads to cystic fibrosis is a deletion of three nucleotides at position 508 in the gene, but several different clinical forms of cystic fibrosis exist – different clinically because they are different genetically ( pubs/cystic/cystic.htm).  And this disease is presented as 'simple'!  In the real world, very little is as simple as your average textbook implies.


Here are the basic “extra complications” with which anyone breeding dogs ought to be familiar:


Incomplete (partial) dominance

            In complete dominance, both GG and Gg (for example) animals are gray – equally gray – and there is no way to tell what their genotypes just by looking at them.  Only gg animals are not gray.  If dominance is incomplete, then GG animals would be gray, gg animals would be non-gray, and Gg animals would be intermediate in phenotype.  What is usually happening in this kind of situation is probably that the dominant allele in a heterozygote Gg is making enough of a functional protein to partially but not fully compensate for the existence of the recessive allele.  (In real life, G appears to be completely dominant to g.)

            An example in the real world of incomplete dominance is seen with the chinchilla dilute which appears to be present in dogs:  CC animals express full pigment; cchcch animals would be cream or white, and Ccch  animals have yellow (phaeomelanin) but not black (eumelanin) pigment diluted.  In horses this gives us the chestnut – palomino – cream (cremello) series of colors.  The chestnut horse has undiluted phaeomelanin, the palomino has some phaeomelanin, and the cream has virtually no phaeomelanin.  In dogs, the chinchilla dilute probably gives us black-and-silver miniature Schnauzers by diluting the tan points on an otherwise black-and-tan dog.  This gene is probably responsible for some of the cream and white breeds, too.

            There is no such thing as a “carrier” for an incompletely dominant trait, as heterozygotes stand out phenotypically from homozygous dominants as well as homozygous recessives.

             Probably there is some incomplete dominance in the spotting series for dogs, such that Ssp animals have more white markings than SS animals, spsw have more white markings that  spsp, and so forth.



            Where incomplete dominance gives us heterozygotes with intermediate phenotype, codominance gives us heterozygotes which express both dominant and recessive traits at the same time.  The classic example is seen in human blood types:  You can be type A (genotype AA), type B (genotype BB) or type AB (genotype AB).  In codominant traits, two different alleles are probably coding for different proteins, both of which are functional, but in different ways.  There are probably quite a few codominant traits, but relatively few which are visible in simple phenotypic terms.



            In this unhappy situation, one genotype is taken out of the picture because it suffers such severe problems that it dies.  Lethals need not be literally all-the-way lethal to have a substantial effect on a breed.  In dogs, merle can act as a deleterious dominant, although not actually lethal in most cases:  Mm animals show merle coloring where they would otherwise have been black, mm animals are non-merle, and MM animals are frequently blind or deaf or both, as well as mostly white.  I've seen sources which indicate that the deleterious effects of the MM genotype seem to be partially or wholly offset if the dog has no other white markings – a MM sheltie is likely to have serious problems, it is said, whereas apparently a MM Dachshund is not.  On the other hand, I've also seen sources which indicate that merle Dachshunds are not free of problems, so I don't know.

            The hairless mutation that creates the special coat variation in Chinese Crested Dogs is a lethal allele.  Breed two hairless Cresteds together and you should get, on average, 1/4 puffs, 1/2 hairless, and 1/4 dead puppies (which I believe are invisible, since they vanish before birth).  Breed a hairless with a puff and you should get 1/2 hairless and 1/2 puffs (and bigger litters, since none are killed by doubling up on the hairless allele).  This is a difficult mutation to talk about genetically because the hairless form is dominant to the powderpuff (normal fully coated form), but the lethality itself acts in a recessive manner -- you need to have two HH (hairless) alleles for the puppy to die.  This kind of trait might be referred to as an incomplete dominant trait with recessive lethality . . . though it would be nice if there was a more concise and clear way to put that!

            The hairless mutation possessed by American Hairless Terriers is a different mutation and not lethal, btw.  Two of the (at least four) pinto patterns seen in horses are lethals.  The "extra" harlequin gene that, in combination with merle, creates the harlequin coat pattern in Great Danes, may also be a lethal (harlequin Danes are probably double heterozygotes:  MmHh).  Breeding around lethals requires basic familiarity with genetics.


Sex linkage (X linkage)

            This simply means that the trait in question is controlled by a gene located on the X chromosome (very few genes are present on the small Y chromosome, other than the ones that say “be male”; this implies that the special attention sometimes paid to the tail-male line in a pedigree is misplaced, since the Y chromosome is the only one which must be passed down this line).  Since dogs have 39 pairs of chromosomes (compared to the human 23), we would not expect a very large percentage of traits to be located on the X chromosome (which implies that the tail-female line in a pedigree is also not particularly important).  However, some forms of hemophilia in dogs (hemophilia A and B) are thought to be X-linked, as is hemophilia in humans, of course.

            If a trait is X-linked and also deleterious, you would expect most of the puppies produced with the trait to be male.  This is because the most typical situation would be a normal but carrier dam bred to a normal sire – both parents would be phenotypically normal, but half the sons (on average) would be affected by the trait (and half the daughters would be carriers).

            If the trait is not significantly deleterious (such as color-blindness in humans), then you would expect affected males to be used in breeding (since no one would care about the trait) and the ratio of affected males and females would probably be closer to even, although even then you would expect more affected males than females.

            The best way to show X-linkage when working with such a trait on paper is probably by putting the gene in question on the X chromosome:  XAXa would be a carrier female; XaY an affected male, for example.  If you do Punnett squares to predict the results of a cross, using this kind of notation keeps the "empty space" on the Y from getting lost.



            Genes are said to be “linked” when they are physically located near one another on the same chromosome.  Linkage becomes important because linked genes do not assort independently.  Linkage is one way you can get normal Mendelian ratios messed up when you do a cross.  You might get double dew claws assorting with white color, for example, even though the traits do not directly influence each other.  (I just made that up; there's no reason to expect that particular linkage to occur).  “Linkage groups” are groups of genes that are all close to one another.  Some genetic tests test directly for genes that are responsible for a problem, but others are linkage tests that test for genes linked to the genes that actually cause the problem.  Linkage is never absolute; it can be broken by crossing over between homologous pairs of chromosomes when gametes (sperm or eggs) are formed.  So genetic tests that depend on linked genes will yield some percentage of false negatives and false positives.  Repeating the test in this case would repeat the false result.  How high a percentage of false results there is depends on how tightly the genes are linked.  The labs that do the tests would know how high a percentage of false results to expect for a particular test.


Epistasis / hypostasis

            Epistasis is a term used to refer to one gene acting “dominant” to a different gene at a different locus.  Note that this is not one allele being dominant to another in the same gene series, but one gene altering the effect of another entirely different gene.  A good example in dogs is the ee genotype at the extension locus creating a clear red or yellow color regardless of what alleles are present at the agouti locus.  Thus Blenheim cavaliers are black-and-tan at the agouti locus, but expression of the black pigment in their coats is suppressed by the red dilute ee.  In this case, the agouti locus would also be said to be hypostatic to the extension locus.

    In high school, generally students are told that brown eyes are dominant to blue, thus leading to the more clueless teachers informing brown-eyed students with both parents blue-eyed that they must have been adopted or the product of adultery.  In fact, human color genetics is not nearly as straightforward as dog color genetics, but the insistence on treating it as though is leads to students being subjected to this stupid and harmful idea by ignorant teachers.  Multiple genes and interactions between them show us how easy it is for brown to be dominant to blue and yet for blue-eyed parents to give rise to brown-eyed children.  For example, suppose that AA or Aa means brown eyes, with aa giving the blue-eyed phenotype.  But in addition, also say that BB or Bb at a separate gene also codes for brown eyes, with bb also giving the blue-eyed phenotype.  So blue is recessive in both cases, right?  But let's say that either aa or bb will give a child blue eyes, regardless of what's going on at the other locus.

    Then this kind of cross, aaBb x Aabb, is between two blue-eyed parents.  But there is a 1/4 chance per child conceived that the child will have brown eyes.  Or say the parents are aaBB x AAbb -- they still both have blue eyes, but now every single one of their children will have brown eyes.

    In fact, something like this system is thought to control blue-vs-brown eyes in humans, with at least one more gene acting to contribute green.  Nobody knows (I think) how you get gray or hazel eyes, so the situation is still more complicated than this in real life.



            When one gene influences more than one trait, it is said to be pleiotropic.  Thus the merle dilute in dogs not only turns black areas into a patchwork of black and gray, it also can influence ear and eye development, particularly if homozygous.  "Extra" effects need not be deleterious.  The cystic fibrosis allele not only causes disease in its homozygous state, but in its heterozygous state also protects against respiratory disease. 



            When many genes influence one trait, the trait is said to be polygenic (or multifactorial).  The two terms are used almost interchangeably in practice, although “multifactorial” is really meant to indicate that part of the influence on the trait comes from non-genetic environmental factors, whereas “polygenic” refers only to genetic factors.  In the real world, it is typical for polygenic traits, such as temperament and hip dysplasia, to show environmental effects.  This is not universal:  the construction of the front assembly in dogs is certainly polygenic, but as far as I know is not thought to be affected by environmental factors.  Usually.  Much.  I mean, horrendously bad nutrition could give some poor creature rickets or otherwise interfere with its growth, which would certainly affect the structure of the front, and that would be environmental influence, obviously. 

            However, a reasonable rule of thumb is:

            If a trait is both complicated anatomically and affected by environmental factors, it is very likely polygenic.  Gastric torsion and bloat is an excellent example of this kind of trait because a tendency to bloat depends on, first, the construction of the chest and abdomen -- complicated anatomically -- and then on the temperament of the dog (shy dogs are more likely to bloat) and whether the food dish is raised (if so, the dog is more likely to bloat).  Although we don't actually know for a fact the mode of inheritance for this problem, it therefore looks an awful lot like a polygenic trait.

            In contrast, if a simple metabolic pathway is easy to visualize for a trait and it does not seem to be affected by environmental factors, it is more likely to be genetically simple – especially if the same or similar traits are known to be simple in other breeds or other species.  Metabolic disorders that depend on the loss of one functional enzyme, such as copper toxicosis, are perfect examples of this.  So is the red color in Cavaliers.  It doesn't much matter what you do with a Cavalier's environment:  red dogs are red and black ones are black.

            Two common models for polygeny are additive polygeny and threshold polygeny.

            Additive polygeny would work like this:  Suppose that each of nine genes adds a “dose” of pigment when present in the dominant form, but not when recessive.  Then an animal with a AABBCcddEEFfGGHHii genotype would have very dark pigment (although not quite as dark as possible), whereas a aabbCcddEEffggHhii animal would have very light pigment (although not as light as possible).  When plotted on a graph, the possible genotypes for this trait would yield a bell-shaped (normal) curve, with most of the animals being intermediate and a “tail” out towards the extremes on both sides.  You could, as a breeder, select for either extreme, but it would not be nearly as easy to achieve consistent dark pigment in this situation as it would if pigment was controlled by a simple one- or two-gene system.

            Traits such as pigment modifiers, height, some kinds of coat texture, and intelligence are easy to visualize as additive traits.  Eye color in dogs is also thought to work this way.  At least one breeder has suggested that in her experience, genes that contribute darker eye pigment are likely to be recessive (, and scroll down to find the comments on eye color -- a lot of the articles at this site are good).

            Threshold polygeny would work like this:  suppose that six genes control the development of hip dysplasia, in such a way that at least three of the genes must be homozygous recessive before any degree of dysplasia exists, and after that more recessive alleles mean the dysplasia is likely to be progressively worse.  (There is no theoretical reason to make recessive genes the bad guys; I'm just doing it that way for the sake of simplicity.

            The three-gene requirement would then be the “threshold” that flips a switch and determines whether dysplasia is present at all; the other alleles plus environmental factors then determine how severe the dysplasia will be.  Thus an animal with a AaBbCcDdEeFf possesses lots of alleles that could contribute to dysplasia, but is itself normal.  It can, and probably will, pass on at least some deleterious alleles to its offspring.  Remember that each gamete gets one allele of each pair -- in the worst case, this dog could make a gamete that contained all recessive (abcdef) alleles, passing on a huge load of deleterious alleles to a puppy (there would be a (1/2)6 chance of it doing so, or a 1/64 chance, which is very small, but it could happen.  In fact, if this was a popular sire and produced, say 64 puppies in one year, then the chances that one puppy would get this combination of alleles would be very high.).

            You can usually assume, when faced with a polygenic trait, that both parents of an affected animal contributed some deleterious alleles.  Their contributions need not, however, have been equal.  This is easy to see with something like the following:

            Dam:  AabbccDdEe - normal                    Sire:  AaBbCcDDEE - normal

                                            Offpring:  aabbccDdEe -- affected (moderately)

            In this case, the dam contributed five deleterious recessive alleles, while the sire only contributed three.

            Models like this seem to have reasonable predictive power when applied to the way many polygenic traits are inherited in dogs.  It's important to assess the siblings of breeding animals in order to make the best guess possible about the genetic quality of the animals you keep to breed, because looking at the whole family makes it easier to assess how many deleterious alleles are likely to be floating around in the family.  More on that in Practical Genetics for the Breeder.


Stuff on the borderline

            A single-gene trait is simple; a polygenic trait is complex.  Sometimes traits fall in between simple and complex.  One form of epilepsy in Welsh Springer Spaniels and Standard Schnauzers is evidently controlled by two genes, both of which must be homozygous recessive for epilepsy to occur – and one of the genes appears to be either on the X chromosome or possibly is part of the mitochondrial DNA (!).  This is why most epileptic animals in these breeds are male.  Probably more traits on the borderline between simple and complex will be discovered as we learn more about the genetic control of problems in dogs.


Trinucleotide repeat disorders (stuttering genes)

            Nucleotides are the basic building blocks of DNA and RNA.  Codons are sets of three nucleotides which code for particular amino acids (which are themselves the building blocks of proteins).  Trinucleotide repeat disorders are caused by the insertion of long sequences of repeated codons in a sequence of DNA.  There are fourteen such disorders known in humans so far.  Of these, Huntington’s disease is one of the best known.  Normally the Huntington’s gene contains about 26 CAG codons.  In people with Huntington’s disease, this number has been expanded to 40 or 100 or even more repeats of this codon.  The same CAG trinucleotide also is involved in eight other known disorders, all of which, like Huntington’s, involve progressive degeneration of nerve cells starting at a relatively advanced age.

            Fragile X syndrome in humans is another trinucleotide repeat disorder, this one involving expansion of a series of CGG repeats on the FMR-1 gene on the X chromosome.

            Trinucleotide repeat disorders frequently are inherited in a weird way.  Sometimes, as in Fragile X, the number of repeats goes up when the affected allele is passed from mother to child, but not when the affected allele is passed from father to child (this may work the other way ‘round for some disorders); and the number of repeats expands sharply once the gene is past a certain number of repeats.  Usually the disorder is worse the more repeats are present -- symptoms are more severe or are expressed earlier in life, usually both.  Thus a trinucleotide repeat disorder might "look" X-linked or Y-linked, even when it’s not. Even weirder, it is likely to increase in severity and penetrance over generations in a family.  It may “look” recessive when it first occurs in a family, but then “look” more and more dominant as generations pass.

            No trinucleotide disorders have been described in dogs, yet, but I wouldn't be surprised if that changed -- and having a picture in the back of your head about what to expect of such a problem might make it easier to spot if such a problem came to light.  Besides, trinucleotide repeats are interesting.  Links below.


Incomplete penetrance

            If all the aa individuals in a population (breed, species) show the aa phenotype, then penetrance is complete and the breeder’s life is easy.  (Easier, anyway.)  If some of the genotypically aa individuals show a A- phenotype, then penetrance is incomplete (and the same, of course, if some A- individuals show the aa phenotype).  If penetrance is low enough, we would start looking for other genes influencing the trait and we would at that point stop thinking of the trait as genetically simple and begin to think of it as probably polygenic or multifactorial.

            Although incomplete penetrance is common, it should not be used as a catch-all explanation for failure of a trait to show expected Mendalian patterns.  Traits that fail to follow Mendalian expectations need to be assessed for other possible complexities, especially polygeny.


Variable expressivity

            Variable expressivity is also common.  This may merely mean that not all affected individuals are affected to the same degree.  Thus some animals with Von Willebrand’s syndrome will be much sicker than others, for example.  In the extreme case, this would grade into incomplete penetrance as animals with the least severe expression of the problem might be clinically normal, even though possessing the "affected" genotype.  Or, possibly, the same underlying genetic system might give rise to syndromes that appear, on the surface, completely different, such as obsessive-compulsive behavior in some individuals and skin disorders in others. Nickolas Dodman suggests in one of his books that a single problem with zinc and/or copper metabolism might give rise to all kinds of problems -- including an awful skin condition called lethal acrodermatitis and a range of obsessive behavior problem -- in bull terriers.  This, too, could be described as variable expressivity.

            In a different way, if what you see on the outside is variable expression of symptoms, this might also indicate that the underlying genetic causation is different for animals that show similar symptoms.  Thus epilepsy created by the two-gene system described above may be found mainly in males, usually first occur between 1 and 3 years of age, and involve infrequent but severe grand mal seizures that are controllable with Phenobarbital; whereas epilepsy created by a different mechanism might be evenly distributed between the sexes, normally show itself between four months and a year of age, and involve frequent but less serious seizures that do not respond well to medication; while yet a third epileptic syndrome might mainly express itself as obsessive “fly-biting” behavior.

            Syndromes which appear grossly similar but vary widely in age of onset, severity, and exact symptoms should probably be suspected of being due to different underlying genetic systems, particularly if the symptoms fall into two or more relatively discrete categories.  On the other hand, syndromes which appear different but could all be affected by a problem with a particular metabolic or developmental pathway should be suspected of being caused by the same underlying genetic factors.



            Many traits which can be created by genetic systems can also be created by non-genetic causes, such as exposure to toxins or traumatic injury.  Thus, epilepsy may be caused by diseases that involve a high fever, such as distemper; exposure to various toxins; or a blow to the head.  In these cases, the epilepsy could correctly be referred to as a phenocopy.  It can be difficult to distinguish phenocopies from genetic problems.  If occurrence of a problem in a breed is rare, sporadic and unpredictable, and especially if the problem does not appear to be associated with any particular line or environment, then it may be due to random environmental or developmental effects and not to any underlying genetic problem.


Monosomies, Trisomies, and Uniparental Disomies

            Usually an individual gets one chromosome of each homologous pair from its mother and one from its father.  An egg cell thus (for dogs) has 39 chromosomes, and so does a sperm cell, and you put the two together and get 78 chromosomes (in 39 pairs).  This is how it's supposed to work.  In the real world, accidents happen.

            If chromosomes do not separate normally during meiotic cell division, by which egg and sperm cells are formed, than you can get a gamete (egg or sperm cell) which does not have the normal number of chromosomes.  This is not normally good for the puppy formed from that gamete.  A monosomy occurs when an individual gets only one chromosome of a pair rather than two (it would have 77 chromosomes total).  In humans, Turner syndrome occurs when an embryo gets one X chromosome from one parent and nothing to match up with it from the other (it is XO rather than XX or XY).  This embryo will grow into a female baby, and the baby will be born alive, but she won't be normal.  She will be sterile, short, with abnormal neck weakness.

            Another possibility, again resulting from an error during meiosis, is that an embryo will get two copies of one chromosome from one parent and one from the other, three copies total.  This is called a trisomy.  In humans, trisomy 21 (three copies of chromosome 21) causes Down syndrome.  There are other viable trisomies, but none as common as Down syndrome.

            A third possible abnormality, even more interesting than either of the above, occurs when an embryo receives two chromosomes from one parent and nothing from the other.  This is called a uniparental disomy, and when it happens there are usually developmental effects, because different genes on a chromosome tend to be activated when you get it from your mother versus when you get it from your father.  Therefore certain genes won't work normally if you get both copies of a chromosome from one parent.  Isn't that interesting?



On to Practical Genetics for the breeder!

Back to the Basics!

*** -- Fragile X -- CGG exon repeat -- Huntington’s Chorea -- CAG exon repeat -- Friedrich’s ataxia -- GAA intron repeat -- trinucleotide repeats in general -- 14 trinucleotide repeat disorders reported in humans so far. -- complexity summarized:  causes of “complexity.” -- linkage analysis of complex trait; Alzheimer’s as a model.