Transmission-ratio distortion and allele sharing in affected sib pairs: a new linkage statistic with reduced bias, with application to chromosome 6q25.3.
A simple multipoint procedure to test for parent-of-origin effects in samples of affected siblings is discussed. The procedure consists of artificially changing all full sibs to half-sibs, with distinct mothers or fathers depending on the parental origin to be evaluated, then analyzing these families with commonly used statistics and software. The procedure leads to tests for linkage through mothers or fathers and also leads to a test for imprinting effects in the presence of linkage. Moreover, simulations illustrate that in regions unlinked to susceptibility genes this multipoint procedure does not have an inflated type I error if a sex-averaged genetic map is used, even when large differences exist between male-specific and female-specific maps. In regions linked with susceptibility genes, the test of imprinting is biased under the null hypothesis if differences exist between sex-specific maps, irrespective of the map used in the analysis. The procedure is applied to the Collaborative Study on the Genetics of Alcoholism dataset from the Genetic Analysis Workshop 14. Results indicate that brothers categorized as affected according to the DMS-III-R and Feighner classification show evidence of linkage through fathers to the 6q25 region (p = 0.00038) as well as modest evidence of imprinting (p = 0.018). This region harbors OPRM1, a candidate gene for substance dependence. Background A simple procedure is proposed to test for linkage through mothers or fathers or to test for imprinting effects in the presence of linkage through the use of nonparametric methods that are based on the estimation of inheritance vectors. This procedure was recently used in an ad hoc fashion, without investigating its properties, to evaluate the excess sharing observed from fathers and mothers in a region of chromosome 6 that was suspected to be under the influence of transmission ratio distortion through mothers, in a sample of families selected through asthmatic probands . The procedure I propose is similar in spirit to what has been described in Paterson et al.  and Karason et al. . Recently, testing the hypothesis of no linkage in samples of affected sib pairs have been shown to gain power by allowing for differences in penetrances depending on the parental origin of the alleles, when such differences indeed exist . In this paper I investigate some properties of the proposed procedure and apply it to the Collaborative Study on the Genetics of Alcoholism (COGA) dataset. from Genetic Analysis Workshop 14: Microsatellite and single-nucleotide polymorphism Noordwijkerhout, The Netherlands, 7-10 September 2004 Published: 30 December 2005 BMC Genetics 2005, 6(Suppl 1):S159 doi:10.1186/1471-2156-6-S1-S159 <supplement> <title> <p>Genetic Analysis Workshop 14: Microsatellite and single-nucleotide polymorphism</p> </title> <editor>Joan E Bailey-Wilson, Laura Almasy, Mariza de Andrade, Julia Bailey, Heike Bickeböller, Heather J Cordell, E Warwick Daw, Lynn Goldin, Ellen L Goode, Courtney GrayMcGuire, Wayne H ning, ail Jarvik, Brion S Maher, Nancy Mendell, Andrew D Paterson, John Rice, Glen Satten, Brian Suar z, Veronica Vieland, Marsha Wilcox, Heping Zhang, Andre s Ziegler and Jean W MacCluer</editor> <note>Proceedings</note> </suppleme t> Page 1 of 5 (page number not for citation purposes) BMC Genetics 2005, 6:S159 Methods The nonparametric Spairs statistic of Whittemore and Halpern  is used as the core statistic to test for linkage. This statistic is transformed into a statistic, Zlr, that is robust against the incompleteness of the marker data, following the method of Kong and Cox . It is asymptotically normally distributed with a mean of 0 and a variance of 1 under the hypothesis of no linkage. Let be the value of the statistic when applied to nuclear families where the sibs have been artificially recoded as half-sibs with distinct mothers. Then the statistic evaluates the excess sharing of alleles from fathers, and is a test of linkage through fathers (more precisely it is a test of Mendelian transmission from fathers to affected sibs). Similarly, let be the value of the statistic when the distinct parents are fathers, which evaluates the sharing of alleles from mothers and is a test of linkage through mothers. I wrote a program, NUCULAR, that splits extended pedigrees into nuclear families with the option of recoding all sibs as half-sibs, with distinct mothers or fathers, who nevertheless maintain their original genetic material. A test of imprinting effects in the presence of linkage can be constructed by evaluating . This statistic has an approximately normal distribution. The mean of the distribution is 0 and the variance is 1 if the region under study is unlinked to any susceptibility gene (due to independent segregation from both parents), but the variance may be unequal to 1 if there is linkage, because the variance of and may be different than 1 and the statistics are likely to be correlated. Consequently, if the region is not linked to any susceptibility genes, the test of imprinting effects has the correct size (but has no power); if the region is linked to a susceptibility gene, the test has power to detect imprinting effects but the true type I error may be different from the nominal one. A solution, if the analysis is carried out using a sexaveraged map, could be a permutation procedure consisting of randomly permuting the genotypes of the parents. This procedure preserves linkage while creating replicates under the null hypothesis of no imprinting effects. It will not correct for the bias introduced if there are large differences in gender-specific genetic maps: the expected values of and may then be different under linkage, due to the reduced power of sparser maps to detect linkage . ALLEGRO v1.2 [3,7] was used to carry out the analyses. To investigate the properties of the tests, multipoint genedropping simulations were carried out using SIMM . In particular, SIMM allows recombination to occur under sex-specific genetic maps. I applied the procedure to the COGA microsatellite dataset, using the DMS-III-R and Feighner classification to define the set of affected individuals. Only one nuclear family was extracted per pedigree, the one with the most affected individuals; this is to allow for a valid permutation procedure as described above. This study was approved by the McGill Institutional Review Board. Results and Discussion The type I error of the proposed methods was evaluated in samples of 250 nuclear families with 2 affected sibs by simulating 4 markers located 5 cM (sex-equal) from one another, each having 4 equally frequent alleles (heterozygosity of 0.75), under the rules of Mendelian inheritance. When creating replicates under linkage, an affection status was modelled based on penetrances of 0.18, 0.48, and Zlr F