A factorial experiment for optimizing the PCR conditions in routine genotyping

Although most PCRs would produce proper PCR products when first tried, a general optimization is required to yield the best results. This optimization is often achieved by changing one factor at a time. However, this may lead to suboptimal results, since interactions between conditions are difficult to detect with this approach. In the present study, we describe the factorial optimization of PCR conditions for microsatellite genotyping, by introducing small systematic variations in conditions during the genotyping process. The hypothesis was that small changes will not affect genotyping results, but will provide information about the optimality of current conditions. The conditions to vary were the concentrations of buffer, MgCl(2), dNTPs, primers, Taq polymerase and DNA, the annealing temperature (T(a)) and the number of cycles. We show that, by a 2(8) factorial experiment, it is possible to identify not only the factors responsible for obtaining good results, but also those responsible for bad results. However, the condition leading to the highest signals is not necessarily the best operational condition. The best operational condition is minimally sensitive to random pipetting fluctuations and yields reliable genotypes as well.

Association testing by haplotype-sharing methods applicable to whole-genome analysis

We propose two new haplotype-sharing methods for identifying disease loci: the haplotype sharing statistic (HSS), which compares length of shared haplotypes between cases and controls, and the CROSS test, which tests whether a case and a control haplotype show less sharing than two random haplotypes. The significance of the HSS is determined using a variance estimate from the theory of U-statistics, whereas the significance of the CROSS test is estimated from a sequential randomization procedure. Both methods are fast and hence practical, even for whole-genome screens with high marker densities. We analyzed data sets of Problems 2 and 3 of Genetic Analysis Workshop 15 and compared HSS and CROSS to conventional association methods. Problem 2 provided a data set of 2300 single-nucleotide polymorphisms (SNPs) in a 10-Mb region of chromosome 18q, which had shown linkage evidence for rheumatoid arthritis. The CROSS test detected a significant association at approximately position 4407 kb. This was supported by single-marker association and HSS. The CROSS test outperformed them both with respect to significance level and signal-to-noise ratio. A 20-kb candidate region could be identified. Problem 3 provided a simulated 10 k SNP data set covering the whole genome. Three known candidate regions for rheumatoid arthritis were detected. Again, the CROSS test gave the most significant results. Furthermore, both the HSS and the CROSS showed better fine-mapping accuracy than straightforward haplotype association. In conclusion, haplotype sharing methods, particularly the CROSS test, show great promise for identifying disease gene loci.

Genetic Association Studies in Complex Disease: Disentangling Additional Predisposing Loci from Associated Neutral Loci Using a Constrained ‐ Permutation Approach

In the process of genetically mapping a complex disease, the question may arise whether a certain polymorphism is the only causal variant in a region. A number of methods can answer this question, but unfortunately these methods are optimal for bi-allelic loci only. We wanted to develop a method that is more suited for multi-allelic loci, such as microsatellite markers. We propose the Additional Disease Loci Test (ADLT): the alleles at an additional locus are permuted within the subsample of haplotypes that have identical alleles at the predisposing locus. The hypothesis being tested is, whether the predisposing locus is the sole factor predisposing to the trait that is in LD with the additional locus under study. We applied ADLT to simulated datasets and a published dataset on Type 1 Diabetes, genotyped for microsatellite markers in the HLA-region. The method showed the expected number of false-positive results in the absence of additional loci, but proved to be more powerful than existing methods in the presence of additional disease loci. ADLT was especially superior in datasets with less LD or with multiple predisposing alleles. We conclude that the ADLT can be useful in identifying additional disease loci.

Efficiency Control in Large-Scale Genotyping Using Analysis of Variance

The efficiency of the genotyping process is determined by many simultaneous factors. In actual genotyping, a production run is often preceded by small-scale experiments to find optimal conditions. We propose to use statistical analysis of production run data as well, to gain insight into factors important for the outcome of genotyping. As an example, we show that analysis of variance (ANOVA) applied to the first-pass results of a genetic study reveals important determinants of genotyping success. The largest factor limiting genotyping appeared to be interindividual variation among DNA samples, explaining 20% of the variance, and a smaller reaction volume, sizing failure, and differences among markers all explained approximately 10%. Other potentially important factors, such as sample position within the plate and reusing electrophoresis matrix, appeared to be of minor influence. About 55% of the total variance could be explained by systematic factors. These results show that ANOVA can provide valuable feedback to improve genotyping efficiency. We propose to adjust genotype production runs using principles of experimental design in order to maximize genotyping efficiency at little additional cost.

Mapping of a susceptibility gene for multiple sclerosis to the 51 kb interval between G511525 and D6S1666 using a new method of haplotype sharing analysis

Multiple sclerosis (MS) is a complex disease that is partly genetic in origin. Although an association of MS with specific human leukocyte antigen (HLA) types has been known for almost 30 years, the nature of this relationship has remained unclear. Furthermore, genetic resolution sufficient to implicate a specific gene in the HLA region has not been achieved. Many loci in the HLA region have been found to be significantly associated with MS, which is largely explained by the extended haplotype sharing and varying marker informativity of the region. We have determined 248 haplotypes of MS patients from the population of the northern Netherlands and 226 haplotypes of their relatives as controls using a set of 22 microsatellite markers covering the HLA region. The data were analyzed using standard association methods and a new statistical method, haplotype sharing statistics (HSS). Haplotype sharing statistics determines the extent of haplotype sharing for all pairs of haplotypes of patients and of controls and calculates the difference in mean haplotype sharing between patients and controls. Haplotype sharing was found to be significantly greater among patients than among controls in a region of 1.1 Mb between markers G511525 and TNFalpha. The involvement of this region is also supported by association analysis and the transmission/disequilibrium test (TDT). Within this region, HSS, which is largely independent of association and TDT, indicated the interval of 51 kb between G511525 and D6S1666 as that most likely to contain a susceptibility gene for MS. As DQB1 is the sole gene known in this interval at present, the results of our analysis suggest that this gene plays a role in the pathogenesis of MS.

Simulation studies of detection of a complex disease in a partially isolated population

Simulation studies were undertaken with POPGEN, a new population simulation program, to explore strategies for detecting loci underlying rare and common disorders in a small population that has been partially isolated for 10 generations. Haplotype-sharing analysis (HSA) and non-parametric linkage analysis (NPL) were applied to the simulated haplotype and pedigree data for 100 cases, 100 controls, and an average of 28 multiplex pedigrees from cases' families, for a 2-5 cM map of markers. When identity by descent (IBD) status was known (using unique founder marker allele designations assigned during simulation), a linkage disequilibrium (LD) signal could be detected under disease-generating models predicting relative risk to sibs of 11.8 (high-RR) or 2.67 (mod-RR). Detection was more difficult when marker alleles were down-coded to resemble microsatellites (heterozygosities 0.75-0.80). False-positive peaks on nondisease chromosomes were uncommon. NPL analysis was more powerful than HSA at this marker density using down-coded alleles and assuming availability of all affected relatives. LD mapping of common disorders is likely to require denser maps of highly polymorphic markers to approximate full IBD information. LD and linkage mapping provide independent information, and strategies that combine these two methods could be useful in studies of small isolated populations.