831
Views
0
CrossRef citations to date
0
Altmetric
Research Articles

A Note on Jöreskog’s ACDE Twin Model: A Solution, Not the Solution

ORCID Icon, ORCID Icon, &
Pages 933-934 | Received 11 May 2022, Accepted 26 Jun 2022, Published online: 20 Jul 2022

Abstract

The classical twin design is used to decompose phenotypic variance into genetic and environmental variance components. In practice, for reasons of identification, either an ADE or an ACE twin model is fitted, i.e., models with three variance components. Jöreskog proposed to use the Moore-Penrose inverse to estimate the four genetic and environmental variance components in the ACDE twin model. We demonstrate that the variance components thus obtained do not equal the true genetic and environmental variance components. Given a ACDE model, resorting to an ADE or ACE model or applying Jöreskog’s method will not produce correct results. However, it is possible that Jöreskog’s method will produce a better approximation to the true variance components under certain configurations of A, D, C, and E variance components.

The univariate classical twin design is a means to the end of decomposing the variance of a given phenotype into the genetic and environmental variance components, based on the phenotypic (2 × 2) monozygotic (MZ) and dizygotic (DZ) covariance matrices, ΣMZ and ΣDZ. On the environmental side, shared or common (C) environmental and unshared (E) environmental variances are distinguished: σC2 and σE2, where C is defined as all non-genetic effects that increase resemblance between relatives (twins). On the genetic side, additive genetic (A) and dominance genetic (D) variance components are distinguished: σA2 and σD2. Subject to many assumptions (Eaves et al., Citation1978), including the independence of the genetic and environmental variables and random mating, the phenotypic variance equals σPh2 = σA2 + σD2+ σC2 + σE2. The observed MZ and DZ (2 × 2) covariance matrices contain three statistics that have distinct model expectations: the phenotypic variance and the MZ and DZ twin covariances. These three statistics are insufficient to resolve all four variance components. As a solution, one may choose, as a point of departure, either the ACE model (σPh2 = σA2 + σC2 + σE2), or the ADE model (σPh2 = σA2 + σD2 + σE2). The choice is based on the following rule of thumb: ACE if rMZ <2*rDZ, and ADE if rMZ >2*rDZ, where rMZ and rDZ are phenotypic correlations. Another solution is to estimate an unsigned variance component for C, and to note that a significant and substantial negative estimate of σC2 implies that variance due to genetic non-additivity outweighs variation due to common environment, assortative mating, or age effects (Verhulst et al., Citation2019). Structural equation modelling (SEM) is used for parameter estimation and model testing, using programs like LISREL (Cardon et al., Citation1991; Jöreskog & Sörbom, Citation2018) or OpenMx (Neale et al., Citation2016). So, from the point of view of SEM, the full ACDE model is not identified. This can be formalized in SEM terms in terms the rank of the information matrix or the rank of the Jacobian (Bollen & Bauldry, Citation2010). Equivalently, it can be formalized terms of the set of equations that related the unknown variance components to the observed information (Jöreskog, Citation2021). In the standardized case (σPh2 = 1), we have σPh2=1=σA2+σD2+σC2+σE2 rMZ=σA2+σD2+σC2 rDZ=½σA2+¼σD2+σC2 where the coefficients ½ and ¼ are based on quantitative genetic theory (Mather & Jinks, Citation1971). This may be expressed as y = Bx, where the column vector xt= [σA2 σD2 σC2 σE2], and the column vector yt = [1, rMZ, rDZ]. The solution involves the inverse is B−1y = x. In the ACDE model the (3 × 4) matrix B has rows [1,1,1,1], [1,1,1,0] and [½, ¼, 1, 0], and the inverse B−1 does not exist. As noted, setting either σC2 to zero (rMZ <2*rDZ) or σD2 to zero (rMZ >2*rDZ) is the common solution to this problem.

Jöreskog (Citation2021) recently proposed an alternative method, namely by using the Moore–Penrose inverse to obtain an estimate of x: x=(Bt(BBt)−1)y. The Moore–Penrose provides the minimum norm solution, i.e., it minimizes √(xtx) given y = Bx. This method can be implemented in programs like LISREL and OpenMx.

The imposition of this mathematical constraint certainly does produce a solution to y = Bx, but we question whether it is the correct quantitative genetic solution. To illustrate this, we consider a phenotype, which is subject to the effect of a single biallelic locus, with two alleles denoted M and N. Let p denote the frequency of allele M (in a well-defined population), and q = 1-p denote the frequency of allele N. Assuming Hardy Weinberg equilibrium, the genotype frequencies are p2 (genotype MM), 2pq (genotype MN) and q2 (genotype NN). The conditional phenotypic means in the subpopulation of individual with genotype MM, MN and NN are µ+a, µ+d, and µ-a, respectively, and the genetic contributions of the locus to the phenotypic variance are (Mather & Jinks, Citation1971; Posthuma et al., Citation2003). σA2= 2pq(a+d(qp))2 σD2=(2pqd)2

For example, given p = 0.5, a = 0.5 and d = a (complete dominance), we have σA2 = 0.125, and σD2 = 0.0625. Suppose furthermore that σC2 = 0.09375 and σE2 = 0.09375, so that σPh2 = 0.375. The standardized values are σAst2 = 4/12, σCst2 = 3/12, σDst2 = 2/12, and σEst2 = 3/12. The MZ and DZ covariance matrices are shown in .

Table 1. MZ and DZ covariance matrices.

The correlations are rMZ = 0.75 and rDZ = 0.4583, which, by the aforementioned rule of thumb, suggests, incorrectly, an ACE model (rMZ <2*rDZ), with estimates, based of Falconer’s equations (Falconer & Mackay, Citation1996), of 2*(rMZ-rDZ) = 0.58333 (σAst2) and 2*rDZ-rMZ = 0.1666 (σCst2). Applying Jöreskog’s method, as implemented in OpenMx, we obtained the estimates shown in .

Table 2. True variance components and estimated variance components based on the Moore–Penrose inverse.

We note that the estimates based on the Moore-Penrose inverse do not equal the true variance components. For instance, the true narrow-sense (σA2Ph2) and broad-sense ((σA2D2)/σPh2) heritabilities are ∼0.3333 (0.125/0.375) and 0.5 ((0.125 + 0.0625)/0.375), respectively. However, the heritabilities based on the results obtained with the Moore-Penrose inverse equal ∼0.2440 (narrow sense) and ∼0.4702 (broad sense).

We conclude that the Moore-Penrose inverse provides a solution for the ACDE model by the imposition of a mathematical constraint. However, it is clear from our results that it does not provide the quantitative genetic solution, i.e., the variance components associated with the genetic and environmental influences. In this connection, we also note that Jöreskog’s Eq. 23, which provides the correct narrow-sense heritabilities in the ACE and ADE model (i.e., σA2), does not do so in the ACDE model. Specifically, according to Eq. 23, and given a standardized phenotype (σPh2 = 1), the narrow sense heritability equals (1/2)*rMZ - (2/7)*rDZ = (5/14)*σAst2+ (3/14)*σCst2 + (6/14)*σDst2, whereas it should equal σAst2.

While we conclude that the solution provided by the Moore–Penrose inverse not the correct (i.e., quantitative genetic) solution, we quickly acknowledge that the common practice of resorting to an ACE or ADE model will not produce the correct results either, if the true model is the ACDE model. In that situation, both solutions produce incorrect results, which may be expressed in terms of an error of approximation. It is likely that, under certain configurations of A, D, C, and E variance components, Jöreskog’s approach has a smaller error of approximation. In closing, we point out that other genetically informative designs have been proposed, in which the four variance components are identified (e.g., Keller et al., Citation2009, Citation2010; Truett et al., Citation1994).

References

  • Bollen, K. A., & Bauldry, S. (2010). Model Identification and computer algebra. Sociological Methods & Research, 39, 127–156. https://doi.org/10.1177/0049124110366238
  • Cardon, L. R., Fulker, D. W., & Jöreskog, K. A. (1991). LISREL 8 model with constrained parameters for twin and adoptive families. Behavior Genetics, 21, 327–350. https://doi.org/10.1007/BF01065971
  • Eaves, L. J., Last, K. A., Young, P. A., & Martin, N. G. (1978). Model-fitting approaches to the analysis of human-behavior. Heredity, 41, 249–320. https://doi.org/10.1038/hdy.1978.101
  • Falconer, D. S., & Mackay, T. F. C. (1996). Introduction to Quantitative Genetics (4th ed.). Pearson.
  • Jöreskog, K. G., & Sörbom, D. (2018). LISREL 10 for Windows [computer software]. Scientific Software International, Inc.
  • Jöreskog, K. G. (2021). Classical models for twin data. Structural Equation Modeling, 28, 121–126. https://doi.org/10.1080/10705511.2020.1789465
  • Keller, M. C., Medland, S. E., Duncan, L. E., Hatemi, P. K., Neale, M. C., Maes, H. H. M., & Eaves, L. J. (2009). Modeling extended twin family data I: Description of the cascade model. Twin Research and Human Genetics, 12, 8–18. https://doi.org/10.1375/twin.12.1.8
  • Keller, M. C., Medland, S. E., & Duncan, L. E. (2010). Are extended twin family designs worth the trouble? A comparison of the bias, precision, and accuracy of parameters estimated in four twin family models. Behavior Genetics, 40, 377–393. https://doi.org/10.1007/s10519-009-9320-x
  • Mather, K., & Jinks, J. L. (1971). Biometrical Genetics (2nd ed.). Cornell University Press.
  • Neale, M. C., Hunter, M. D., Pritikin, J. N., Zahery, M., Brick, T. R., Kirkpatrick, R. M., Estabrook, R., Bates, T. C., Maes, H. H., & Boker, S. M. (2016). OpenMx 2.0: Extended Structural Equation and Statistical Modeling. Psychometrika, 81, 535–549. https://doi.org/10.1007/s11336-014-9435-8
  • Posthuma, D., Beem, A. L., de Geus, E. J., van Baal, G. C., von Hjelmborg, J. B., Iachine, I., & Boomsma, D. I. (2003). Theory and practice in quantitative genetics. Twin Research, 6, 361–376. https://doi.org/10.1375/136905203770326367
  • Truett, K. R., Eaves, L. J., Walters, E. E., Heath, A. C., Hewitt, J. K., Meyer, J. M., Silberg, J., Neale, M. C., Martin, N. G., & Kendler, K. S. (1994). A model system for analysis of family resemblance in extended kinships of twins. Behavior Genetics, 24, 35–49. https://doi.org/10.1007/BF01067927
  • Verhulst, B., Prom-Wormley, E., Keller, M., Medland, S., & Neale, M. C. (2019). Type I error rates and parameter bias in multivariate behavioral genetic models. Behavior Genetics, 49, 99–111. https://doi.org/10.1007/s10519-018-9942-y