HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Maniatis, N. Articles by Pollott, G. E. Search for Related Content PubMed PubMed Citation Articles by Maniatis, N. Articles by Pollott, G. E.

The impact of data structure on genetic (co)variance components of early growth in sheep, estimated using an animal model with maternal effects

Imperial College of Science, Technology and Medicine, Department of Agricultural Sciences, Wye, Ashford, Kent, TN25 5AH, U.K.

² Correspondence:
phone: +44 (0)20 759 42707; fax: +44 (0)20 759 42919; E-mail:
g.pollott{at}ic.ac.uk.

	Abstract

Top Abstract Introduction Material and Methods Results Discussion Implications Literature Cited

 
Several studies have noted high negative correlations between maternal genetic and direct additive effects and their influence on additive and maternal heritability of early growth traits in sheep. Multigeneration data from the Suffolk Sire Reference Scheme (SSRS) were used to investigate the effect of data structure on estimates of direct and maternal (co)variances for lamb 8-wk weight. In all analyses the additive, maternal genetic, maternal environmental, and residual effects were fitted along with the covariance between direct and maternal additive effects. The contributions of particular genetic relationships to the estimates were studied by analyzing subsets of the SSRS data. A further eight subsets were formed having 10% or 50% of the dams with their own records and having one or two, three or four, five or six, and more than six offspring per dam. Analysis of data having only 10% of the dams with their own record and one or two offspring records yielded a high negative correlation (-0.99) between direct and maternal genetic effects. However, the seven other data sets with more records per dam or a higher proportion of dams with their own records produced values of -0.35 to -0.51. Data structure and the number of dams and granddams with records are important determinants of estimated direct and maternal effects in early growth traits.

Key Words: Data Analysis • Genetic Parameters • Growth • Lambs • Maternal Effects

	Introduction

Top Abstract Introduction Material and Methods Results Discussion Implications Literature Cited

 
Selection schemes for early growth and carcass traits that ignore maternal effects are likely to produce suboptimal genetic responses (Gerstmayr, 1992). Genetic models, including maternal effects and the covariance of direct and maternal genetic effects, fit data better than the simple additive model. High negative correlations between direct and maternal genetic effects for early weight traits are common but biologically impossible. Therefore, the interpretation of results is not straightforward, reflecting the difficulty in separating these effects from the permanent environmental components, when all are estimated in the same analysis (Maniatis and Pollott, 2002a,b,c). One possible reason for large negative estimates for the direct-maternal correlation is the structure of the data (Meyer, 1992a). The animal model utilizes all relationships available in a given data set. The direct additive genetic component is estimated from a single observation per animal, whereas maternal effects depend on the number of progeny per dam, the number of dams with recorded performance (Gerstmayr, 1992), and the number of generations of recorded data.

Numerous studies have investigated the importance of maternal effects in a mixed-model setting for various livestock species. The information on the impact of data structure using the current modelling methodology is scarce. Only a small number of studies have investigated the impact of data structure on parameter estimates in a model with maternal effects (e.g., Baker, 1980; Gerstmayr, 1992; Meyer, 1992b), and these were based solely on simulated data sets. The objective of this work was to study the effect of particular genetic relationships, the number of offspring per dam, and the influence of the proportion of dams with recorded performance on parameter estimates, using farm records.

	Material and Methods

Top Abstract Introduction Material and Methods Results Discussion Implications Literature Cited

 
Data
Data from the Suffolk Sire Reference Scheme Ltd. (SSRS), previously described by Maniatis and Pollott (2002b), were analyzed in the present study. Matings take place in late summer with the use of artificial insemination and therefore, genetic connectedness is achieved by distributing reference sires across members’ flocks. The present study was based on an analysis of 8-wk BW, which was obtained from 58 Suffolk flocks participating in the SSRS between 1985 and 1997. This weight was calculated by linear interpolation between birth weight and a weight taken at about 8-wk of age. Lambing year within flock was used as the basis for contemporary groups. Recent studies have shown the importance of increased prediction error variance when a large number of groups having limited number of records per subclass are used (Van Vleck, 1987; Tosh and Wilton, 1994; Van Bebber et al., 1997). The 501 contemporary groups in the present study were large with an average number of progeny per subclass of 111. Tosh and Kemp (1994) demonstrated that the prediction error variance was minimized when records for animals in contemporary groups of five or less were excluded. In the present study, contemporary groups that had less than 10 animals were omitted and contemporary group was treated as a fixed effect. There were 15,244 dams in the data with average number of progeny records per dam of 3.65. Seventy percent of the dams had fewer than eight progeny with records, of which 27% had four to six lambs.

Statistical Analyses
Direct additive genetic (a), maternal additive genetic (m), and maternal environmental (c) effects were considered by including the appropriate random effects in the model. The full linear mixed model was as follows:

where y, b, a, m, c, and e are vectors of observations, fixed effects, direct additive genetic effects, maternal additive genetic effects, permanent maternal environmental effects, and random environmental residual effects, respectively. The fixed effect part of the model comprised flock/year of record, sex (male, female), type of birth/rearing (seven levels), and age of the dam (1 to 7 and >7 yr of age), as described by Maniatis and Pollott (2002b). Incidence matrices ZA, Z_M, and Z_C relate the observations to the respective fixed and random effect. The variance and covariance structure for the random effects in the analyses can be described as

where A, I, I_m, ²_a, ²_m, _am, ²_c, and ²_e, are the numerator relationship matrix, identity matrix for animals, identity matrix for dams, direct additive genetic variance, maternal additive genetic variance, direct-maternal covariance, maternal environmental, and residual error variance, respectively. This model and numerator relationship matrix was fitted to all data sets used in the study. In some analyses a simple additive model was also fitted for comparative purposes.

Estimation of Variance Covariance Components
All calculations were carried out using the restricted maximum likelihood method (Patterson and Thompson, 1971) based on a derivative-free algorithm (DFREML program of Meyer, 1998). Fitting an animal model with several random effects, the Simplex procedure was employed to locate the maximum of the log likelihood (logL) as described by Meyer (1998). For each trait in the analysis convergence was considered to have been reached when the variance of function values (-2*logL) was less than 10^-8.

All the (co)variance components, with phenotypic variance (²_p) being the sum of ²_a, ²_m, _am, ²_c, and ²_e, were derived at convergence. The direct-maternal genetic correlation (r_am) and the total heritability (h²_T) were calculated after convergence. Total-additive variance used in the total heritability calculation was defined by Willham (1972) as ²_{total-additive} = (²_a + 0.5 ²_m + 1.5 _am). The estimated (co)variance components were used to obtain direct heritability (h²), maternal heritability (m²), the covariance between direct and maternal effects as a proportion of the phenotypic variance (c_AM), maternal permanent environmental variance as a proportion of ²_p (c²), and direct-maternal additive genetic correlation (Standard errors were calculated for all variance ratios after Smith and Graser (1986).

Constructing Specific Types of Data Sets
The first population structure (PS1) was formed from dam-offspring record pairs but without records on any maternal granddams. The paternal granddam-offspring pairs were retained in the data since its covariance does not contribute to the maternal additive genetic variance and the direct-maternal covariance estimations (Willham, 1963; Thompson, 1976; Meyer, 1992a).

A number of studies have postulated the possibility of a negative environmental covariance between dam and offspring. A negative environmental correlation between a mother’s early growth and her subsequent milking ability has been suggested (Koch, 1972; Baker, 1980), which implies that daughters of dams with very good mothering ability may provide an inferior environment for their offspring. By summarizing literature results, Baker (1980) argued that, when estimating direct-maternal genetic correlations for birth weight until weaning, it is important to exclude dam-offspring relationships because they could be biased due to the negative environmental covariance. Therefore, another data set was created (PS2) where dam-offspring record pairs were excluded from the analysis but all granddams with recorded performance were retained.

The relationship between family structure and sampling (co)variances was investigated by Meyer (1992b), based on a list of covariances between relatives that contain information on direct and maternal (co)variances given by Thompson (1976). She simulated six different family structures and demonstrated that covariances between grandparents and grand offspring provide information for the estimation of the direct-maternal genetic covariance. Thus, an additional subset of the data (PS3) was created, which included records from animals and their mothers and their maternal granddam, with a record. The covariances between relatives that arise in the three populations (PS1, PS2, and PS3) are shown in Table 1.

The pedigree structure is shown in Table 2 for the 55,683 lambs, the offspring of 1,804 sires. Mean 8-wk weight was 24.8 ± 5.10 kg with a coefficient of variation of 20.5%.

	Results

Top Abstract Introduction Material and Methods Results Discussion Implications Literature Cited

One feature of Table 2 is the differences in the direct-maternal additive genetic correlations that reflected the differences between the first two family structures (PS1 and PS2) vs PS3, which had a much higher proportion of recorded dams (83%) and granddams (70%) than the full data set (60% and 42%, respectively). The PS1 set with approximately 60% of the dams having records, excluded records on maternal grandparents (i.e., two generations in total). The analysis yielded a negative direct-maternal additive genetic correlation of -0.44, which was more negative than both PS3 and the full data set, even though the latter had a similar proportion of dams with their own record (57%). This change in r_am ^{was not associated with large changes in h2}, m², and c² when compared with the full and PS3 results. It also yielded the smallest value of total heritability (12%), which was approximately half that estimated by the simple additive model (23%). The level of covariance between the direct and maternal genetic values was low in all analyses, being less than 6% in all cases.

With only offspring-maternal grandparent record relationships in the data (PS2), the analysis yielded a considerably higher negative estimate of r_am of -0.60, accompanied by a higher estimate of ²_a with a corresponding estimate of h² of 0.24. Excluding all dam-offspring record pairs in the PS2 data set yielded an m² estimate of 0.03 (i.e., the m² parameter was largely estimated from the pedigree links to the dams). This reduced value reflected the difficulty in disentangling the genetic components using the PS2 data set. Furthermore, the maternal permanent environmental effect increased by 27% over that obtained from the PS1 and PS3 population types. With h² explaining 24% of the total variation, the h²_T parameter was higher (18%) compared to PS1 but with an analogous difference in the direct heritability from the simple additive model.

The PS3 data set included animals that had a dam and at least one maternal granddam with recorded performance, with a number of families also having great-granddams or more female ancestors with records (i.e., > three generations). Compared to PS1, the number of dams in PS3 with their own record was higher. As noted for other relatives, the data also included maternal granddams and 70% of them had their own record. The PS3 data also contained a higher proportion of sire-offspring and grand sire-offspring relationships relative to PS1. These differences in data structure were reflected in the direct-maternal correlation, which was found to be considerably lower than PS1 and PS2. With h² the only exception, parameter estimates were found to be independent of the structure of PS3 since estimates agreed very closely with those obtained by analyzing the full data set.

The Effect of Number of Offspring per Dam and Proportion of Dams with Recorded Performance
Tables 3 and 4 give estimates of the genetic parameters for different sizes of dam progeny group when 10% (Group A; Table 3) or 50% (Group B; Table 4) of the dams with recorded performance were included. The striking results from Table 3 demonstrate substantial differences in parameter estimates across population types. The analysis of PT1A yielded the most unfavorable results compared to population types PT2A to PT4A. The negative correlation between direct and maternal additive genetic effects was found to be large (-0.99), even though a sizeable data set was used (10,187 records). In PT1A there was a reduction in m² and an increase in c². Thus, the change in correlation has arisen more as a consequence of the decreases of the direct and maternal additive genetic variances rather than as a reduction in the covariance. With 43% of the dams having one or two offspring records, the maternal heritability explained only 4% of the total variation and with 6,556 maternal half-sibs in the data the permanent environment effect due to the dam explained a large portion of the total variation (28%). As a result the total heritability was minimal (0.05).

By further increasing the number of progeny up to five or six per dam and > six (PT3A and PT4A), the number of maternal half-sib subclasses decreased compared to PT2A, and as a result both direct and maternal heritabilities were reduced considerably. For PT4A, estimates of h² and m² decreased by approximately 40% (over PT2A). This is because larger maternal half-sib combinations imply a larger number of relationships among individuals whose mothers are related (e.g., maternal cousins). Although PT1A had even more maternal half-sibs, its structure, as previously discussed, did not allow accurate estimation of all parameters.

The results in Table 4 show that the effect of the proportion of recorded dams on parameter estimation was of similar importance to that of the number of progeny per dam. With 50% of dams having records, the direct-maternal correlations decreased in magnitude compared with those obtained from families that have had only 10% (Table 3). Results from the first population type showed a substantial increase of the maternal heritability (from PT1A to PT1B) accompanied by a much less negative r_am (from -0.99 to -0.51) and consequently a smaller c². It was only in the second population type, where the number of dams with records was not of primary significance, that no major differences were observed in PT2B compared with PT2A. Analyzing the two subsequent types (PT3B and PT4B) resulted in noticeable decreases of m², by approximately 30% in both cases, and much less negative values for r_am compared with PT3A and PT4A, respectively.

Examining the effect of number of offspring per dam, in population type B, it is clearly shown that the magnitude of negative r_am declined with increasing number of progeny records, and this decline was found to be more consistent than estimates derived from PT1A to PT4A. The most unfavorable case was again PT1B that yielded the strongest negative r_am of -0.51 and a maternal effect that appeared to be more of environmental that genetic in origin. In analyses that followed (PT2B to PT4B) the value of r_am became increasingly smaller by approximately 12% at each population type until it reached the lowest value of -0.35, and this pattern was accompanied analogously with decreases in maternal heritability estimates.

	Discussion

Top Abstract Introduction Material and Methods Results Discussion Implications Literature Cited

 
The Impact of Population Structure
Three data structures were analyzed differing in mother-offspring, granddam-offspring relationships in the data. The PS1 data set included only records from dams and their offspring, while for PS2 all dams with records were omitted but with all remaining granddam-offspring record relationships included in the data. The PS3 data set included all recorded dams and granddams. Conceptually, the PS3 data set included all dams of the dams, since the 8-wk BW data consisted of 13 yr of records (e.g., great-granddams) until the earliest maternal ancestor with recorded performance.

Contrasting the results obtained by analyzing the full and PS1 data sets, the change in the genetic direct-maternal correlation was large enough to consider the important contribution of incorporating recorded maternal granddams (-0.36 and -0.44, respectively). In addition, PS2 yielded a smaller estimate of m² (3%), since there was no maternal genetic effect that could be passed from mother to offspring. Consequently, the reduced estimates of m² increased the value of h² (24%) compared to the full and PS1 data sets.

Willham (1972) modeled the granddam effects based on the idea that the dam’s early environment (as a daughter) influences her subsequent performance as a mother. This implies that the granddam of the offspring would exert a direct effect through the mother on the phenotypic value of her daughter’s offspring. Baker (1980) suggested that inclusion of grand-offspring records could increase (absolute value) the negative estimates of r_am considerably. However, the fact that a large number of dams will be missing in the data could have serious consequences on the direct-maternal covariance, as previously illustrated.

Negative estimates for the direct-maternal correlation still persisted in the PS3 data set. The results presented from these three data sets (PS1 to PS3) should be regarded with some caution since emphasis was given only to offspring-dam-granddam-great granddam combinations (e.g., other relationships between relatives in the data were ignored) in both constructing the data sets and interpreting the results. Hagger and Schneeberger (1995) concluded that the estimation of direct-maternal covariance from field data is a delicate task even with informative combinations of relatives and manageable data sets. These authors reported larger negative correlations than the ones presented in this study.

The Impact of Data Structure
The great variation in parameter estimates that was observed by analyzing different population types, with the main emphasis on the differences in maternal heritability and direct-maternal correlation, demonstrated the consequences of a deficient structure on the estimation of maternal effects. Based on the findings presented here, there is strong evidence to suggest that both the number of progeny per dam and the proportion of mothers with recorded performance considerably influence the parameter estimates. The present results confirm the findings of Gerstmayr (1992). Based on simulated data, the author showed that the accuracy of estimation of maternal effects depends on the family structure and demonstrated that the proportion of dams having their own record in the data and the number of progeny had an important influence on (co)variance components. With large negative correlations the differences between population structures were more pronounced. Analyzing extreme structures yielded strong negative correlations and therefore showed the apparent difficulty of disentangling the direct effect into maternal (co)variance components.

With a small size of progeny group per dam and limited information from recorded dams the direct-maternal correlation had the highest (negative) value. In setting up the design matrices that relate records through the dam to their maternal effects, estimates of the maternal additive genetic component are for all animals in the analysis while estimates of c² are only for dams with progeny records. Therefore, with only an average of 1.5 offspring per dam the incidence matrices Z_M and Z_C tend to be similar. This causes difficulties in separating maternal additive and maternal environmental effects, which depends on the magnitude of the covariances between the dam effects relative to their variances. The small maternal heritability and the large estimate for c² (0.28) could also verify this. Keeping the same size of progeny group but increasing the number of dams with recorded performance produced a considerably lower negative correlation. Earlier studies have indicated that the estimation of maternal effects and their covariance components is inherently problematic since the expression of those effects lag by one generation (Willham, 1980). Maternal inheritance is always expressed with a time lag of one generation; hence, a genetic lag of 0.5 represents the coefficient of dam-offspring relationship. Therefore, having sufficient dam-offspring record pairs will improve the separation of the maternal effects.

The size of the maternal heritability coincided with the direct-maternal correlation for the first and second types of progeny groups. By increasing the number of offspring the correlation became smaller (negative). This may have been caused by the increased number of animals per maternal half-sib group with their covariance being a quarter of the additive genetic variance (i.e., only the dams as common parent), and unity for direct-maternal covariance. Similar pattern of results was also observed by Gerstmayr (1992).

A number of studies (Meyer, 1992b; Robinson, 1994; Hagger and Schneeberger, 1995; Robinson 1996a,b; Maniatis and Pollott, 2002a, b) that have identified strong antagonistic interactions between direct and maternal effects have observed large increases of both the direct and maternal additive genetic variances. The general concept reported from the above studies is that large negative correlations will result in increases of the direct and maternal heritabilities in order to compensate for such strong negative correlations. According to the present results this is not the case. Analyzing the PT1A data set yielded m² of 0.04 with r_am of -0.99. It is obvious that limited data are the primary cause of large negative, and biologically impossible, direct-maternal correlations.

The subsets PT1 and PT2 (one to two or three to four offspring per dam), of both Groups A and B (10% or 50% of dams with recorded performance), had a higher number of maternal half-sibs than data PT3B (five or six offspring of 50% of recorded dams). Larger maternal half-sib combinations imply larger number of relationships among individuals whose mothers are related (e.g., maternal cousins). According to Willham (1963), these relationships reduce the environmental contribution to the maternal variance. Perhaps the increase of c² in data sets PT2B and PT3B affected the maternal heritability, which was found to be smaller in PT3B.

	Implications

Top Abstract Introduction Material and Methods Results Discussion Implications Literature Cited

 
The estimation of maternal effects and the correlation between direct and maternal genetic effects is dependent on key pedigree relationships. A high proportion of both dams and maternal granddams with their own records is essential. The separation between the maternal genetic component and permanent maternal environmental effects requires repeated records on individual mothers and the presence of these mothers in the data. Thus, several well-linked generations of records and many relationships between relatives related to the mother are needed. The main problem with field data sets is that often the types of covariances between relatives available in the data do not have sufficiently different expectations to allow the accurate partition of the phenotypic variance into all maternal (co)variance components. One of the key benefits of sire-reference schemes is the large half-sib groups of dams that are formed by the extensive use of the reference sires.

	Footnotes

 
¹ This work was undertaken while the senior author was on a studentship funded by the Hellenic State Scholarship Foundation. The Suffolk Sire Referencing Scheme Ltd. provided the data used in this study. The help from both organizations is gratefully acknowledged.

Received for publication March 11, 2002. Accepted for publication September 27, 2002.

	Literature Cited

Top Abstract Introduction Material and Methods Results Discussion Implications Literature Cited

 


Baker, R. L. 1980. The role of maternal effects on the efficiency of selection in beef cattle: a review. Proc. N. Z. Soc. Anim. Prod. 40:285–303.

Gerstmayr, S. 1992. Impact of the data structure on the reliability of the estimated genetic parameters in an animal model with maternal effects. J. Anim. Breed. Genet. 109:321–336.

Hagger, C., and M. Schneeberger. 1995. Influences of amount of pedigree information on computing time and of model assumptions on restricted maximum-likelihood estimation of population parameters in Swiss Black-Brown Mountain sheep. J. Anim. Sci. 73:2213–2219.[Abstract]

Koch, R. M. 1972. The role of maternal effects in animal breeding. VI. Maternal effects in beef cattle. J. Anim. Sci. 35:1316–1323.

Maniatis, N., and G. E. Pollott. 2002a. Direct and maternal (co)variances for weight and ultrasonically measured traits of lambs in a closed Suffolk flock. Small Ruminant Res. 45:235–246.

Maniatis, N., and G. E. Pollott. 2002b. Nuclear, cytoplasmic and environmental effects on growth, fat and muscle traits in Suffolk lambs from a sire referencing scheme. J. Anim. Sci. 80:57–67.[Abstract/Free Full Text]

Maniatis, N., and G. E. Pollott. 2002c. Genotype-by-environment interactions in lamb weight and composition traits. Anim. Sci. 75:3–14.

Meyer, K. 1992a. Bias and sampling covariances of estimates of variance components due to maternal effects. Genet. Sel. Evol. 24:487–509.

Meyer, K. 1992b. Variance components due to direct and maternal effects for growth traits of Australian beef cattle. Livest. Prod. Sci. 31:179–204.

Meyer, K. 1998. DFREML. Version 3a, User notes, Animal Breeding and Genetics Unit., University of New England, Armidale, NSW, Australia.

Patterson, L. D., and R. Thompson. 1971. Recovery of inter-block information when block sizes are unequal, Biometrika 58:545–554.[Abstract/Free Full Text]

Robinson, D. L. 1994. Models which might explain negative correlations between direct and maternal genetic effects. In Proc. 5th World Cong. Genet. Appl. Livest. Prod., Guelph, Canada 18:378–381.

Robinson, D. L. 1996a. Estimation and interpretation of direct and maternal genetic parameters for weights of Australian Angus cattle. Livest. Prod. Sci. 45:1–11.

Robinson, D. L. 1996b. Models which might explain negative correlations between direct and maternal genetic effects. Livest. Prod. Sci. 45:111–122.

Smith, S. P., and H.-U. Graser. 1986. Estimating variance components in a class of mixed models by restricted maximum likelihood. J. Dairy Sci. 69:1156–1165.[Abstract/Free Full Text]

Thompson, R. 1976. The estimation of maternal genetic variances. Biometrics 32:903–917.[Medline]

Tosh, J. J., and R. A. Kemp. 1994. Estimation of variance components for lamb weights in three sheep populations. J. Anim. Sci. 72:1184–1190.[Abstract]

Tosh, J. J., and J. W. Wilton. 1994. Effects of data structure on variance of prediction error and accuracy of genetic evaluation. J. Anim. Sci. 72:2568–2577.[Abstract]

Van Bebber, J., N. Reinsch, W. Junge, and E. Kalm. 1997. Accounting for herd, year and season effects in genetic evaluations of dairy cattle: a review. Livest. Prod. Sci. 51:191–203.

Van Vleck, L. D. 1987. Contemporary groups for genetic evaluations. J. Dairy Sci. 70:2456–2464.

Willham, R. L. 1963. The covariance between relatives for characters composed of components contributed by related individuals. Biometrics 19:18–27.

Willham, R. L. 1972. The role of maternal effects in animal breeding: III. Biometrical aspects of maternal effects in animals. J. Anim. Sci. 35:1288–1324.

Willham, R. L. 1980. Problems in estimating maternal effects. Livest. Prod. Sci. 7:405–418.

This article has been cited by other articles:

				A. Molina, A. Menendez-Buxadera, M. Valera, and J. M. Serradilla Random regression model of growth during the first three months of age in Spanish Merino sheep J Anim Sci, November 1, 2007; 85(11): 2830 - 2839. [Abstract] [Full Text] [PDF]

				G. Banos, S. Brotherstone, and M. P. Coffey Prenatal Maternal Effects on Body Condition Score, Female Fertility, and Milk Yield of Dairy Cows J Dairy Sci, July 1, 2007; 90(7): 3490 - 3499. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Maniatis, N. Articles by Pollott, G. E. Search for Related Content PubMed PubMed Citation Articles by Maniatis, N. Articles by Pollott, G. E.