Epistasis and phantom heritability

A few years ago, I proposed the "lego-block" model of genetic variation. The main idea of this model is that the search for genes that "do" something is often misguided, because most genes don't actually do much of anything in themselves: they are commodity blocks (like lego pieces), and it is the way they are put together that influences the complex phenotypic outcome of an organism.

(By not "doing" something, I do not, of course, mean that they have no biological effect. Rather, I mean that they have no effect at a higher-order phenotypic trait, in the same way that the luminosity of pixels is irrelevant to the depicted image, it is rather the combination of pixels of different luminosity that produces an image)

I am not, of course, denying that there are genes with large positive/negative effects, but these traverse one of two possible trajectories during evolution:

• Flicker: alleles of large negative effect arise, may persist for a few generations, but ultimately die out. They never amount to much of anything
• Shine: alleles of large positive effect spread through the population quickly and become fixed

Alleles that flicker are much more common than alleles that shine. This is a simple consequence of the random nature of mutation: there are many more ways to break a system than to improve it, and the mutation mechanism is excellent in breaking down organisms: the survivors are the ones who carry a "manageable" mutation load, and, once in a very long while, adaptive alleles arise that quickly become fixed.

An alternative to the "lego-block" model of commodity alleles that produce positive/negative phenotypes due to the way they work together (epistasis) is the model of additive variation. According to this model, there is a plethora of genes of small positive/negative effect for a trait, and the final phenotypic expression is influenced by the sum of positive/negative alleles one inherits from their parents.

A new paper in PNAS provides strong evidence that the missing heritability is not due to our inability of finding loci of small effect, but rather to the fact that we've overestimated the heritability of traits. 

At the limit (a perfect "lego-block" world) there are absolutely no alleles that are individually associated with any traits. That does not mean that all individuals will be phenotypically indistinguishable! A great deal of phenotypic variation can still persist even in this case.

Here is a simple example:

	C	D
A	10	0
B	0	10

The value of a trait depending on the alleles in two loci is shown, e.g., AC=10, AD=0, BC=0, BD=10.

It can be easily seen (due to symmetry) that whether one inherits A/B in one locus, or C/D in the other has no effect -in itself- on the trait. It is the combination of alleles that has a (huge) effect on the trait.

The paper is open access.

PNAS doi: 10.1073/pnas.1119675109

The mystery of missing heritability: Genetic interactions create phantom heritability

Or Zuk et al.

Human genetics has been haunted by the mystery of “missing heritability” of common traits. Although studies have discovered >1,200 variants associated with common diseases and traits, these variants typically appear to explain only a minority of the heritability. The proportion of heritability explained by a set of variants is the ratio of (i) the heritability due to these variants (numerator), estimated directly from their observed effects, to (ii) the total heritability (denominator), inferred indirectly from population data. The prevailing view has been that the explanation for missing heritability lies in the numerator—that is, in as-yet undiscovered variants. While many variants surely remain to be found, we show here that a substantial portion of missing heritability could arise from overestimation of the denominator, creating “phantom heritability.” Specifically, (i) estimates of total heritability implicitly assume the trait involves no genetic interactions (epistasis) among loci; (ii) this assumption is not justified, because models with interactions are also consistent with observable data; and (iii) under such models, the total heritability may be much smaller and thus the proportion of heritability explained much larger. For example, 80% of the currently missing heritability for Crohn's disease could be due to genetic interactions, if the disease involves interaction among three pathways. In short, missing heritability need not directly correspond to missing variants, because current estimates of total heritability may be significantly inflated by genetic interactions. Finally, we describe a method for estimating heritability from isolated populations that is not inflated by genetic interactions.

Link

Half of hidden heritability found (for height, at least)

This is a quite interesting paper, as it shows, by sampling a large number of individuals), that the heritability of height is not missing after all. The authors looked at a large number of individuals, and this allowed them to discover statitically significant associations between height and more SNPs than before.

This bears great promise as it may hint that genome-wide association studies, that have come under substantial criticism lately, may be failing not because of an inherent flaw, but rather because they are not sampling enough individuals.

The discovered SNPs account for 45% of the heritability of height. Where is the rest? The authors argue for two additional sources:

First, SNPs in current microarray chips sample the genome incompletely. Locations in-between discovered SNPs are in incomplete linkage disequilibrium with the discovered SNPs. So, there is undetected polymorphism, in the gaps between the hundreds of thousands of SNPs in current chips, that may explain a portion of the missing heritability.

Second, SNPs have different minor allele frequencies. For example, in one SNP the minor allele may occur at 10% of individuals, while in others at 30%. This is important, because it is more difficult to arrive at a statistically significant result in the former case.

Consider a SNP with a minor allele frequency of 2%. Then, if you sample 1,000 individuals, only about 20 of them are expected to have the minor allele. You cannot estimate the average height of the minor allele with a sample of 20 people as securely as you can with a sample of 500. Thus, if the SNP influences height in a small way, you will not be able to detect it.

A further complication, which I've written about before, is that some variation in the human genome is family-related, or at least occurs at fewer individuals than the allele frequency cutoff. If 99.9% of people have C at a given location and 0.1% of people have T, this variant is unlikely to be included in a microarray chp, because it is too rare to matter economically: you would only get a handful of individuals -if you're lucky- in a sample of 1,000 for such a variant. However, rarity does not mean that the variant is functionally unimportant, and the rare allele may play a substantial role in the height of the people who possess it.

The publication of this paper is a cause for optimism, as it shows that progress can be made by brute force: fuller genome coverage and more individuals. We'll have to wait and see whether or not the same approach will work for other complex traits, such as IQ or schizophrenia, that have been hitherto difficult to crack.

Obviously, the cost of sampling more individuals will become an issue in future studies, but the cost-per-individual is expected to drop. So, I'm guessing that more discoveries are in store for us in the next few years.

UPDATE (Jun 28):

Not the main point of the paper, but also included in the supplementary material (pdf) are some nice PCA results.

In the European-only PCA we see the familiar north-south gradient (anchored by Tuscans TSI and Netherlands NET on either side), and the orthogonal deviation of the Finns. Swedes (SWE) occupy a northern European end of the spectrum like the Dutch, but are spread towards Finns, reflecting low-level Finnish admixture in that population. Conversely, Finns are variable along the same axis, reflecting variable levels of admixture. Australians (AUS) and UK, on the other hand, are on the northern European edge of the main European gradient, with a number of individuals spread toward the Tuscan side.

The PCA with all populations is also quite interesting. East Eurasians (Chinese and Japanese) form a tight pole at the bottom right. Gujarati Indians (GIH) form a different pole, spread towards Europeans, reflecting variable levels of West Eurasian admixture in that population, probably corresponding to the ANI element recently discovered in Indian populations. Mexicans (MEX) are spread towards East Asians, reflecting their Amerindian admixture, but notice how they are not positioned exactly on the European-East Asian axis, probably reflecting the third, minority, Sub-Saharan element in their ancestry, as well as the fact that Amerindians are not perfectly represented by East Asians. Finns are tilted towards East Asians, as expected, reflecting the fact that their genetic specificity vis a vis Northern Europeans is due to low-level East Eurasian ancestry.

An interesting aspect of the first two PCs is the fact that the Maasai (MKK) and Luhya (LUW) from Kenya are not separated from Caucasoids, and neither are Yoruba from Nigeria (YRI). This is a good reminder of the fact that identity in the first two principal components may mask difference revealed in higher order components. This difference (at least for Maasai) is seen in the next two PCs.

Nature Genetics doi:10.1038/ng.608

Common SNPs explain a large proportion of the heritability for human height

Jian Yang et al.

Abstract

SNPs discovered by genome-wide association studies (GWASs) account for only a small fraction of the genetic variation of complex traits in human populations. Where is the remaining heritability? We estimated the proportion of variance for human height explained by 294,831 SNPs genotyped on 3,925 unrelated individuals using a linear model analysis, and validated the estimation method with simulations based on the observed genotype data. We show that 45% of variance can be explained by considering all SNPs simultaneously. Thus, most of the heritability is not missing but has not previously been detected because the individual effects are too small to pass stringent significance tests. We provide evidence that the remaining heritability is due to incomplete linkage disequilibrium between causal variants and genotyped SNPs, exacerbated by causal variants having lower minor allele frequency than the SNPs explored to date.

Link

Why Some Women Look Young for Their Age (Gunn et al. 2009)

On the left facial composites of younger/older (left/right) monozygotic/dizygotic twins (top/bottom).

PLoS ONE doi:10.1371/journal.pone.0008021

Why Some Women Look Young for Their Age

David A. Gunn et al.

Abstract

The desire of many to look young for their age has led to the establishment of a large cosmetics industry. However, the features of appearance that primarily determine how old women look for their age and whether genetic or environmental factors predominately influence such features are largely unknown. We studied the facial appearance of 102 pairs of female Danish twins aged 59 to 81 as well as 162 British females aged 45 to 75. Skin wrinkling, hair graying and lip height were significantly and independently associated with how old the women looked for their age. The appearance of facial sun-damage was also found to be significantly correlated to how old women look for their age and was primarily due to its commonality with the appearance of skin wrinkles. There was also considerable variation in the perceived age data that was unaccounted for. Composite facial images created from women who looked young or old for their age indicated that the structure of subcutaneous tissue was partly responsible. Heritability analyses of the appearance features revealed that perceived age, pigmented age spots, skin wrinkles and the appearance of sun-damage were influenced more or less equally by genetic and environmental factors. Hair graying, recession of hair from the forehead and lip height were influenced mainly by genetic factors whereas environmental factors influenced hair thinning. These findings indicate that women who look young for their age have large lips, avoid sun-exposure and possess genetic factors that protect against the development of gray hair and skin wrinkles. The findings also demonstrate that perceived age is a better biomarker of skin, hair and facial aging than chronological age.

Link

Genetics and environment contributions to craniofacial phenotypes of Belgians

Hum Biol. 2008 Dec;80(6):637-54.

Contribution of genetics and environment to craniofacial anthropometric phenotypes in Belgian nuclear families.

Jelenkovic A, Poveda A, Susanne C, Rebato E.

In this study we estimate relative genetic and environmental influences on head-related anthropometric phenotypes. The subject group consisted of 119 nuclear families living in Brussels, Belgium, and included 238 males and 236 females, ages 17 to 72 years. Two factor analyses with varimax rotation (the first one related to facial measurements and the second one to overall head morphology) were used to analyze 14 craniofacial size traits. The resulting four synthetic traits [HFCF, VFCF, HDF1, and HDF2-horizontal (breadth) and vertical (height) facial factors and two head horizontal (breadth) factors, respectively] were used as summary variables. Maximum heritabilities (H2) were estimated for all studied traits, and variance components analysis was applied to determine the contribution of genetics and environment on the four craniofacial factors. In addition, we examined the covariations between the face (HFCF and VFCF) and head-related factors (HDF1 and HDF2), separately. Quantitative genetic analysis showed that HFCF, VFCF, HDF1, and HDF2 variation was appreciably attributable to additive genetic effects, with heritability (h2) estimates of 67.62%, 54.97%, 70.76%, and 65.05%, respectively. The three variance components reflecting a shared familial environment were nonsignificant for these four phenotypes. Bivariate analysis revealed significant additive and residual correlations for both pair of traits. The results confirm the existence of a significant genetic component determining the four craniofacial synthetic traits, and common genetic and environmental effects shared by the two face-related phenotypes and by the head-related ones.

Link

Common variants and schizophrenia

Until now, the hunt for the genetic etiology of psychiatric disorders didn't go very well. Three new papers in Nature make some progress by noting that a combination of many genes, especially of the Major Histocompatibility Complex is predictive of schizophrenia and bipolar disorder risk. From the NIH news release:

Three schizophrenia genetics research consortia, each funded in part by NIMH, report separately on their genome-wide association studies online July 1, 2009, in the journal Nature. However, the SGENE, International Schizophrenia (ISC) and Molecular Genetics of Schizophrenia (MGS) consortia shared their results – making possible meta-analyses of a combined sample totaling 8,014 cases and 19,090 controls.

All three studies implicate an area of Chromosome 6 (6p22.1), which is known to harbor genes involved in immunity and controlling how and when genes turn on and off. This hotspot of association might help to explain how environmental factors affect risk for schizophrenia. For example, there are hints of autoimmune involvement in schizophrenia, such as evidence that offspring of mothers with influenza while pregnant have a higher risk of developing the illness.

From news.com.au:

As well as pinpointing key immune system mutations, complementary discoveries from each consortium showed clearly that many small genetic variations combine in different ways to increase a person's risk of developing schizophrenia.

"If you look at any individual with schizophrenia no single gene is really strong, but put these genes together and you get a meaningful influence," Dr Cairns said.

From the deCODE news release:

"Genetics offers a unique window for better understanding diseases like schizophrenia because the brain and cognition are so little understood and so difficult to study. Discoveries such as these are crucial for teasing out the biology of the disease and making it possible for us to begin to develop drugs targeting the underlying causes and not just the symptoms of the disease. One of the reasons this study was so successful is its unprecendented size. Pooling our resources has yielded spectacular results, which is what the participants from three continents hoped for. At the same time, this study underscores the fact that rare variants may well carry a significant part of the genetic risk of schizophrenia, so our next task is to use the ever more affordable sequencing technologies to find more of them," said Kari Stefansson, CEO of deCODE and corresponding author on the paper.

In the first phase of the study, the deCODE-led SGENE consortium conducted a genome-wide scan of more than 300,000 SNPs in a total of 17,000 patients and controls from England, Finland, Germany, Iceland, Italy and Scotland. The 1500 SNPs with the best signal were then analyzed in 11,000 patients and controls from the International Schizophrenia Consortium (ISC) and the European-American portion of the Molecular Genetics of Schizophrenia studies (MGS). Twenty-five SNPs with strong suggestive correlation were then followed up in more than 20,000 patients and controls from the Netherlands, Denmark, Germany, Hungary, Norway, Russia, Finland and Spain. Bringing together the results of different consortia established he association between the the total of seven markers on chromosomes 6, 11, and 18 with increased risk of schizophrenia.

So, while these three studies are a vindication of sorts for common variants, the greater part of the risk remains to be found in rare variants that are not captured by the 300K SNPs or so that were genotyped. But, by any means, this is a significant victory in the search for the hidden heritability.

From the Stanford release:

Using commercially available "SNP chips" designed to detect those more-common variants, the investigators looked for differences between the DNA of people with schizophrenia versus the DNA of those without the disease. The scientists required that such differences achieve "genome-wide statistical significance." Here's why: If you flip a million coins, one at a time, you're going to see all kinds of seemingly miraculous events — say, 15 heads in a row — that may seem significant but are typical when you toss even a perfectly balanced coin so many times.

Shi's job was to devise analytical techniques to determine whether the "finding" of a SNP's greater likelihood among schizophrenics was real or spurious. The genomic region on chromosome 6 survived this rigorous statistical test.

"These findings show that our genetic methods are working, and that the genetic underpinnings of schizophrenia can be understood," said Levinson. "Similar methods have produced critical new discoveries in many other common diseases, once very large numbers of people could be studied. Now we see that the same approach works for psychiatric disorders like schizophrenia."

The three papers (in no particular order):

Nature doi:10.1038/nature08185

Common polygenic variation contributes to risk of schizophrenia and bipolar disorder

The International Schizophrenia Consortium

Abstract

Schizophrenia is a severe mental disorder with a lifetime risk of about 1%, characterized by hallucinations, delusions and cognitive deficits, with heritability estimated at up to 80%1, 2. We performed a genome-wide association study of 3,322 European individuals with schizophrenia and 3,587 controls. Here we show, using two analytic approaches, the extent to which common genetic variation underlies the risk of schizophrenia. First, we implicate the major histocompatibility complex. Second, we provide molecular genetic evidence for a substantial polygenic component to the risk of schizophrenia involving thousands of common alleles of very small effect. We show that this component also contributes to the risk of bipolar disorder, but not to several non-psychiatric diseases.

Link

Nature doi:10.1038/nature08186

Common variants conferring risk of schizophrenia

Hreinn Stefansson et al.

Abstract

Schizophrenia is a complex disorder, caused by both genetic and environmental factors and their interactions. Research on pathogenesis has traditionally focused on neurotransmitter systems in the brain, particularly those involving dopamine. Schizophrenia has been considered a separate disease for over a century, but in the absence of clear biological markers, diagnosis has historically been based on signs and symptoms. A fundamental message emerging from genome-wide association studies of copy number variations (CNVs) associated with the disease is that its genetic basis does not necessarily conform to classical nosological disease boundaries. Certain CNVs confer not only high relative risk of schizophrenia but also of other psychiatric disorders1, 2, 3. The structural variations associated with schizophrenia can involve several genes and the phenotypic syndromes, or the 'genomic disorders', have not yet been characterized4. Single nucleotide polymorphism (SNP)-based genome-wide association studies with the potential to implicate individual genes in complex diseases may reveal underlying biological pathways. Here we combined SNP data from several large genome-wide scans and followed up the most significant association signals. We found significant association with several markers spanning the major histocompatibility complex (MHC) region on chromosome 6p21.3-22.1, a marker located upstream of the neurogranin gene (NRGN) on 11q24.2 and a marker in intron four of transcription factor 4 (TCF4) on 18q21.2. Our findings implicating the MHC region are consistent with an immune component to schizophrenia risk, whereas the association with NRGN and TCF4 points to perturbation of pathways involved in brain development, memory and cognition.

Link

Nature doi:10.1038/nature08192

Common variants on chromosome 6p22.1 are associated with schizophrenia

Jianxin Shi et al.

Abstract

Schizophrenia, a devastating psychiatric disorder, has a prevalence of 0.5–1%, with high heritability (80–85%) and complex transmission1. Recent studies implicate rare, large, high-penetrance copy number variants in some cases2, but the genes or biological mechanisms that underlie susceptibility are not known. Here we show that schizophrenia is significantly associated with single nucleotide polymorphisms (SNPs) in the extended major histocompatibility complex region on chromosome 6. We carried out a genome-wide association study of common SNPs in the Molecular Genetics of Schizophrenia (MGS) case-control sample, and then a meta-analysis of data from the MGS, International Schizophrenia Consortium and SGENE data sets. No MGS finding achieved genome-wide statistical significance. In the meta-analysis of European-ancestry subjects (8,008 cases, 19,077 controls), significant association with schizophrenia was observed in a region of linkage disequilibrium on chromosome 6p22.1 (P = 9.54 times 10-9). This region includes a histone gene cluster and several immunity-related genes—possibly implicating aetiological mechanisms involving chromatin modification, transcriptional regulation, autoimmunity and/or infection. These results demonstrate that common schizophrenia susceptibility alleles can be detected. The characterization of these signals will suggest important directions for research on susceptibility mechanisms.

Link

19th century trumps 21st (for predicting height)

What this study has found is that predicting a person's height from that of his parents is much more accurate than predicting it from all discovered height-related genetic loci combined.

Of course, we need to pursue genomics, since that is the only way in which we will eventually learn how height is inherited, and which biological factors affect growth. But, for practical purposes, and for the foreseeable future, I doubt that we'll get more information about a baby's prospects by looking at its genes, than by looking at its family.

For some background on the underwhelming results of genomics see In search of the hidden heritability.

European Journal of Human Genetics advance online publication 18 February 2009; doi: 10.1038/ejhg.2009.5

Predicting human height by Victorian and genomic methods

Yurii S Aulchenko et al.

Abstract

In the Victorian era, Sir Francis Galton showed that |[lsquo]|when dealing with the transmission of stature from parents to children, the average height of the two parents, |[hellip]| is all we need care to know about them|[rsquo]| (1886). One hundred and twenty-two years after Galton's work was published, 54 loci showing strong statistical evidence for association to human height were described, providing us with potential genomic means of human height prediction. In a population-based study of 5748 people, we find that a 54-loci genomic profile explained 4-6% of the sex- and age-adjusted height variance, and had limited ability to discriminate tall|[sol]|short people, as characterized by the area under the receiver-operating characteristic curve (AUC). In a family-based study of 550 people, with both parents having height measurements, we find that the Galtonian mid-parental prediction method explained 40% of the sex- and age-adjusted height variance, and showed high discriminative accuracy. We have also explored how much variance a genomic profile should explain to reach certain AUC values. For highly heritable traits such as height, we conclude that in applications in which parental phenotypic information is available (eg, medicine), the Victorian Galton's method will long stay unsurpassed, in terms of both discriminative accuracy and costs. For less heritable traits, and in situations in which parental information is not available (eg, forensics), genomic methods may provide an alternative, given that the variants determining an essential proportion of the trait's variation can be identified.

Link

Heritability of Human cranial dimensions

J Anat. 2009 Jan;214(1):19-35.

Heritability of human cranial dimensions: comparing the evolvability of different cranial regions.

Martínez-Abadías N, Esparza M, Sjøvold T, González-José R, Santos M, Hernández M.

Quantitative craniometrical traits have been successfully incorporated into population genetic methods to provide insight into human population structure. However, little is known about the degree of genetic and non-genetic influences on the phenotypic expression of functionally based traits. Many studies have assessed the heritability of craniofacial traits, but complex patterns of correlation among traits have been disregarded. This is a pitfall as the human skull is strongly integrated. Here we reconsider the evolutionary potential of craniometric traits by assessing their heritability values as well as their patterns of genetic and phenotypic correlation using a large pedigree-structured skull series from Hallstatt (Austria). The sample includes 355 complete adult skulls that have been analysed using 3D geometric morphometric techniques. Heritability estimates for 58 cranial linear distances were computed using maximum likelihood methods. These distances were assigned to the main functional and developmental regions of the skull. Results showed that the human skull has substantial amounts of genetic variation, and a t-test showed that there are no statistically significant differences among the heritabilities of facial, neurocranial and basal dimensions. However, skull evolvability is limited by complex patterns of genetic correlation. Phenotypic and genetic patterns of correlation are consistent but do not support traditional hypotheses of integration of the human shape, showing that the classification between brachy- and dolicephalic skulls is not grounded on the genetic level. Here we support previous findings in the mouse cranium and provide empirical evidence that covariation between the maximum widths of the main developmental regions of the skull is the dominant factor of integration in the human skull.

Link

Epigenetics via twin studies

Coverage elsewhere:

Epigenetics reveals unexpected, and some identical, results
Inherited traits may explain differences in 'identical' twins

Related: In search of the Hidden Heritability

Nature Genetics doi: 10.1038/ng.286

DNA methylation profiles in monozygotic and dizygotic twins

Zachary A Kaminsky et al.

Abstract

Twin studies have provided the basis for genetic and epidemiological studies in human complex traits. As epigenetic factors can contribute to phenotypic outcomes, we conducted a DNA methylation analysis in white blood cells (WBC), buccal epithelial cells and gut biopsies of 114 monozygotic (MZ) twins as well as WBC and buccal epithelial cells of 80 dizygotic (DZ) twins using 12K CpG island microarrays. Here we provide the first annotation of epigenetic metastability of 6,000 unique genomic regions in MZ twins. An intraclass correlation (ICC)-based comparison of matched MZ and DZ twins showed significantly higher epigenetic difference in buccal cells of DZ co-twins (P = 1.2 10^-294). Although such higher epigenetic discordance in DZ twins can result from DNA sequence differences, our in silico SNP analyses and animal studies favor the hypothesis that it is due to epigenomic differences in the zygotes, suggesting that molecular mechanisms of heritability may not be limited to DNA sequence differences.

Link

Genetic predisposition for fathering sons

A good explanation from BBC News:

Newcastle University researchers found men were more likely to have sons if they had more brothers and vice versa if they had more sisters.

They looked at 927 family trees, with details on 556,387 people from North America and Europe, going back to 1600.

The same link between sibling sex and offspring sex was not found for women.

...

In the years after World War I, there was an upsurge in boy births, and Dr Gellatly said that a genetic shift could explain this.

The odds, he said, would favour fathers with more sons - each carrying the "boy" gene - having a son return from war alive, compared with fathers who had more daughters, who might see their only son killed in action.

However, this would mean that more boys would be fathered in the following generation, he said.

Evolutionary Biology doi:10.1007/s11692-008-9046-3

Trends in Population Sex Ratios May be Explained by Changes in the Frequencies of Polymorphic Alleles of a Sex Ratio Gene

Corry Gellatly

Abstract

A test for heritability of the sex ratio in human genealogical data is reported here, with the finding that there is significant heritability of the parental sex ratio by male, but not female offspring. A population genetic model was used to examine the hypothesis that this is the result of an autosomal gene with polymorphic alleles, which affects the sex ratio of offspring through the male reproductive system. The model simulations show that an equilibrium sex ratio may be maintained by frequency dependent selection acting on the heritable variation provided by the gene. It is also shown that increased mortality of pre-reproductive males causes an increase in male births in following generations, which explains why increases in the sex ratio have been seen after wars, also why higher infant and juvenile mortality of males may be the cause of the male-bias typically seen in the human primary sex ratio. It is concluded that various trends seen in population sex ratios are the result of changes in the relative frequencies of the polymorphic alleles of the proposed gene. It is argued that this occurs by common inheritance and that parental resource expenditure per sex of offspring is not a factor in the heritability of sex ratio variation.

Link

In search of the Hidden Heritability

Nature has a very interesting high level survey of the problem of the "hidden heritability". While many traits such as height, autism, or schizophrenia are known to be significantly heritable, recent genome scans with high-density microarray chips, that look at hundreds of thousands of DNA polymorphisms, have failed to produce any significant results.

So, if these traits are in our genes, how come we can't find them there?

The article does a great job at identifying the possible ways to find the "hidden heritability". Here they are, in my own words:

1. Look at more DNA spots

There is a long way between the million or so DNA bases covered by current microarray chips and the whole human genome. Because of linkage disequilibrium, i.e., DNA's propensity to be cut and inherited in large chunks, and not small pieces, you can often tell the value of a marker by looking at nearby markers. But, still, you don't really know until you look. So, denser microarrays, or even whole genome sequences may uncover some of the hidden heritability.

2. Look at more people

Associations between traits and genes are established by statistics. To find a weak association, or an association between a not-so-common variant and the trait in question, you need a large sample. So, if the hidden heredity is hidden away in markers that are beyond your statistical power, you can simply increase this power: sample more people.

3. Look at copy-number variations (CNVs)

Any two individuals don't just have single-letter differences, but also structural changes, where an individual may have more or fewer copies of entire chunks of DNA. So, by looking at single nucleotides you are examining one source of human variation, but missing another chunk of it that may as important.

3. Study gene-gene interactions

Genes form complex networks of interaction. If you flip a SNP from C to T, you don't always get the same effect on the phenotype. This flip may increase, decrease, or leave unaffected, your risk for a disease, depending on what other genes you have. This epistatic interaction of genes makes it difficult to detect associations. It's a lot easier to study the individual effects of 2N alleles at N genes than it is to study the effects of 2^N possible combinations.

4. Don't trust heritability estimates

What if inherited conditions thought to be genetic aren't really genetic, because of epigenetic modifications of gene expression, or shared environments (e.g., in the womb) that aren't accounted for?

5. Don't trust diagnoses of conditions

If you want to find a correlation between a gene G and a trait T, you'd better be sure what T actually is. If it's a whole set of different behaviors, conveniently bundled into a condition T (such as schizophrenia), then you're in trouble, since each of these conditions may have its own causative agent. Many major diseases may be caused by more than one underlying condition, with a different genetic background. So, if you are seeking to find the common thread between people with trait T, you might not find it because there is no common thread!

My guess is that the bulk of the missing heritability is to be found in three sources:

1. Epistasis. Humans are makeshift accidents of evolution, and not well-engineered machines where the effects of individual components have been designed to work well in isolation, shielding other components from their effects. Most things in the human body affects most other things, either directly or indirectly. There are, of course, some master switches which do have individual pronounced effects (e.g., giving you lactose tolerance or breast cancer), but these are the exception. Normal variation is due to how well-put together the individual is, and not so much in the individual components.
2. Gene-Environment interactions. Just as the effect of genes depends on the joint presence of other genes (epistasis), so it depends on the presence of particular environmental influences. Imagine an allele that shows zero association with a particular trait. Does this mean that it has no influence on that trait? No, since zero association is perfectly compatible with even a huge influence, provided that a positive influence under one type of genomic or environmental background is balanced by a negative influence under another.
3. Very low frequency (family) alleles. Natural selection faces a constant battle against the continuing re-emergence of less-than-optimal alleles. Children are almost certainly on average genetically worse than their parents, since parents have survived and reproduced, while children's ability to do so is yet to be tested (*) While human variation is -in part- due to long-lived alleles that have braved the generations, quite a lot of it is due to recent alleles that arose in families, and have not had the time to spread to many bodies. It is these extremely rare family alleles and allele combinations that population studies can't quite capture.
   

Read the original story at Nature: Personal genomes: The case of the missing heritability.

Some related posts on the limits of genome-wide association studies: on intelligence, on height and body mass index, and on CNVs.

(*) Incidentally, this is why the population replacement rate is more than 2 children per woman.