Is bad luck the main cause of cancer?

A recent study reports that the log lifetime incidence rate across a selection of 31 cancer types is highly correlated with the log of the estimated tissue-specific lifetime number of stem cell divisions. This observation, which underscores the importance of errors in DNA replication, has been viewed as implying that most cancers arise through unavoidable bad luck, leading to the suggestion that research efforts should focus on early detection, rather than etiology or prevention. We argue that three statistical issues can, if ignored, lead analysts to incorrect conclusions. Statistics for traffic fatalities across the United States provide an example to demonstrate those inferential pitfalls. While the contribution of random cellular events to disease is often underappreciated, the role of chance is necessarily difficult to quantify. The conclusion that most cases of cancer are fundamentally unpreventable because they are the result of chance is unwarranted.

Optimally weighted Z-test is a powerful method for combining probabilities in meta-analysis.

The inverse normal and Fisher's methods are two common approaches for combining P-values. Whitlock demonstrated that a weighted version of the inverse normal method, or 'weighted Z-test', is superior to Fisher's method for combining P-values for one-sided T-tests. The problem with Fisher's method is that it does not take advantage of weighting and loses power to the weighted Z-test when studies are differently sized. This issue was recently revisited by Chen, who observed that Lancaster's variation of Fisher's method had higher power than the weighted Z-test. Nevertheless, the weighted Z-test has comparable power to Lancaster's method when its weights are set to square roots of sample sizes. Power can be further improved when additional information is available. Although there is no single approach that is the best in every situation, the weighted Z-test enjoys certain properties that make it an appealing choice as a combination method for meta-analysis.

[Fate of parental mitochondria in embryonic stem hybrid cells].

When hybrid cells are created, not only nuclear genomes of parental cells unite but their cytoplasm as well. Mitochondrial DNA (mtDNA) is a convenient marker of cytoplasm allowing one to gain insight into the organization of hybrid cell cytoplasm. We analyzed the parental mtDNAs in hybrid cells resulting from fusion of Mus musculus embryonic stem (ES) cells with splenocytes and fetal fibroblasts of DD/c mice or with splenocytes of M. caroli. Identification of the parental mtDNAs in hybrid cells was based on polymorphism among the parental mtDNAs for certain restrictases. We found that intra- and inter-specific ES cell-splenocyte hybrid cells lost entirely or partially mtDNA derived from the somatic partner, whereas ES cell-fibroblast hybrids retained mtDNAs from both parents in similar ratios with a slight bias. The lost of the "somatic" mitochondria by Es-splenocyte hybrids implies non-random segregation of the parental mitochondria as supported by a computer simulation of genetic drift. In contrast, ES cell-fibroblast hybrids show bilateral random segregation of the parental mitochondria judging from analysis of mtDNA in single cells. Preferential segregation of "somatic" mitochondria does not depend on the differences in sequences of the parental mtDNAs but depends on replicative state of the parental cells.

Use of pairwise marker combination and recursive partitioning in a pharmacogenetic genome-wide scan.

The objective of pharmacogenetic research is to identify a genetic marker, or a set of genetic markers, that can predict how a given person will respond to a given medicine. To search for such marker combinations that are predictive of adverse drug events, we have developed and applied two complementary methods to a pharmacogenetic study of the hypersensitivity reaction (HSR) associated with treatment with abacavir, a medicine that is used to treat HIV-infected patients. Our results show that both of these methods can be used to uncover potentially useful predictive marker combinations. The pairwise marker combination method yielded a collection of marker pairs that featured a spectrum of sensitivities and specificities. Recursive partitioning results led to the genetic delineation of multiple risk categories, including those with extremely high and extremely low risk of HSR. These methods can be readily applied in pharmacogenetic candidate gene studies as well as in genome-wide scans.

Truncated product method for combining P-values.

We present a new procedure for combining P-values from a set of L hypothesis tests. Our procedure is to take the product of only those P-values less than some specified cut-off value and to evaluate the probability of such a product, or a smaller value, under the overall hypothesis that all L hypotheses are true. We give an explicit formulation for this P-value, and find by simulation that it can provide high power for detecting departures from the overall hypothesis. We extend the procedure to situations when tests are not independent. We present both real and simulated examples where the method is especially useful. These include exploratory analyses when L is large, such as genome-wide scans for marker-trait associations and meta-analytic applications that combine information from published studies, with potential for dealing with the "publication bias" phenomenon. Once the overall hypothesis is rejected, an adjustment procedure with strong family-wise error protection is available for smaller subsets of hypotheses, down to the individual tests.

Evolution of the simulated data problem.

The simulated data problem was designed via an interactive process by the Simulation Problem Organizing Committee and the selected data simulators. Based on discussions at the previous Genetic Analysis Workshop, many of the features of previous simulation problems, such as a complex disease, genome scan, and replication, were retained and in addition, a population genetics model was used to generate the simulated genes. We describe the process that was used to structure the problem and summarize the discussions about many of the scientific issues that were considered.

GAW12: simulated genome scan, sequence, and family data for a common disease.

The Genetic Analysis Workshop (GAW) 12 simulated data involves a common disease defined by imposing a threshold on a quantitative liability distribution. Associated with the disease are five quantitative risk factors, a quantitative environmental exposure, and a dichotomous environmental variable. Age at disease onset and household membership were also simulated. Genotype data, including 2,855 microsatellites on 22 autosomes, were simulated for 1,497 individuals in 23 families. Phenotype data and sequence data for seven candidate genes were provided for 1,000 of these individuals who were "living" and available for study. Data were simulated for 50 replicate samples in each of two populations, a general population and a population isolate formed from a small group of founders.

Identifying susceptibility genes using linkage and linkage disequilibrium analysis in large pedigrees.

Linkage and linkage disequilibrium tests are powerful tools for mapping complex disease genes. We investigated two approaches to identifying markers associated with disease. One method applied linkage analysis and then linkage disequilibrium tests to markers within linked regions. The other method looked for linkage disequilibrium with disease using all markers. Additionally, we investigated using Simes' test to combine p-values from linkage disequilibrium tests for nearby markers. We applied both approaches to all replicates of the Genetic Analysis Workshop 12 problem 2 isolated population data set. We reported results from the 25th replicate as if it were a real problem and assessed the power of our methods using all replicates. Using all replicates, we found that testing all markers for linkage disequilibrium with disease was more powerful than identifying markers that were in linkage with disease and then testing markers within those regions for linkage disequilibrium with the implementations that we chose. Using Simes' test to combine p-values for linkage disequilibrium tests on correlated markers seemed to be of marginal value.

Association mapping: where we've been, where we're going.

This paper provides a review of recent work in the area of marker-phenotype association studies, specifically as used for localizing--or mapping--genes affecting a trait of interest. We describe the basis of association mapping and discuss a number of the commonly used techniques. We have also included references to various papers that have evaluated the use of these methods.

Novel tests for marker-disease association using the Collaborative Study on the Genetics of Alcoholism data.

We applied several novel tests for association and linkage in the presence of association to the Genetic Analysis Workshop 11 Problem 1 data set. Our analyses included a Hardy-Weinberg test for association between a marker and a disease susceptibility locus, a Bayesian transmission/disequilibrium test, and a Bayesian case-control test. Positive results for each of these methods require the presence of population association between the marker and a disease susceptibility locus.

On the potential for estimating the effective number of breeders from heterozygote-excess in progeny.

The important parameter of effective population size is rarely estimable directly from demographic data. Indirect estimates of effective population size may be made from genetic data such as temporal variation of allelic frequencies or linkage disequilibrium in cohorts. We suggest here that an indirect estimate of the effective number of breeders might be based on the excess of heterozygosity expected in a cohort of progeny produced by a limited number of males and females. In computer simulations, heterozygote excesses for 30 unlinked loci having various numbers of alleles and allele-frequency profiles were obtained for cohorts produced by samples of breeders drawn form an age-structured population and having known variance in reproductive success and effective number. The 95% confidence limits around the estimate contained the true effective population size in 70 of 72 trials and the Spearman rank correlation of estimated and actual values was 0.991. An estimate based on the heterozygote excess might have certain advantages over the previous estimates, requiring only single-locus and single-cohort data, but the sampling error among individuals and the effect of departures from random union of gametes still need to be explored.

Exact tests for association between alleles at arbitrary numbers of loci.

Associations between allelic frequencies, within and between loci, can be tested for with an exact test. The probability of the set of multi-locus genotypes in a sample, conditional on the allelic counts, is calculated from multinomial theory under the hypothesis of no association. Alleles are then permuted and the conditional probability calculated for the permuted genotypic array. The proportion of arrays no more probable than the original sample provides the significance level for the test. An algorithm is provided for counting genotypes efficiently in the arrays, and the powers of the test presented for various kinds of association. The powers for the case when associations are generated by admixture of several populations suggest that exact tests are capable of detecting levels of association that would affect forensic calculations to a significant extent.