ABSTRACT
Interval-specific congenic strains (ISCS) allow fine mapping of a quantitative trait locus (QTL), narrowing its confidence interval by an order of magnitude or more. In earlier work, we mapped four QTL specifying differential ethanol sensitivity, assessed by loss of righting reflex because of ethanol (LORE), in the inbred long-sleep (ILS) and inbred short-sleep (ISS) strains, accounting for approximately 50% of the genetic variance for this trait. Subsequently, we generated reciprocal congenic strains in which each full QTL interval from ILS was bred onto the ISS background and vice versa. An earlier paper reported construction and results of the ISCS on the ISS background; here, we describe this process and report results on the ILS background. We developed multiple ISCS for each Lore QTL in which the QTL interval was broken into a number of smaller intervals. For each of the four QTL regions (chromosomes 1, 2, 11 and 15), we were successful in reducing the intervals significantly. Multiple, positive strains were overlapped to generate a single, reduced interval. Subsequently, this reduced region was overlaid on previous reductions from the ISS background congenics, resulting in substantial reductions in all QTL regions by approximately 75% from the initial mapping study. Genes with sequence or expression polymorphisms in the reduced intervals are potential candidates; evidence for these is presented. Genetic background effects can be important in detection of single QTL; combining this information with the generation of congenics on both backgrounds, as described here, is a powerful approach for fine mapping QTL.
Subject(s)
Alcohol-Induced Disorders, Nervous System/genetics , Genetic Predisposition to Disease/genetics , Quantitative Trait Loci/genetics , Sleep Wake Disorders/genetics , Alcohol-Induced Disorders, Nervous System/metabolism , Alcohol-Induced Disorders, Nervous System/physiopathology , Animals , Brain Chemistry/genetics , Chromosome Mapping/methods , DNA Mutational Analysis/methods , Drug Resistance/genetics , Female , Genetic Testing/methods , Genotype , Male , Mice , Mice, Neurologic Mutants , Mutation/genetics , Polymorphism, Genetic/genetics , Sleep/genetics , Sleep Wake Disorders/metabolism , Sleep Wake Disorders/physiopathology , Species SpecificityABSTRACT
Quantitative trait locus (QTL) mapping is often done in a single segregating population, such as a backcross or an intercross. Both QTL location and effect size are then estimated from the same dataset. This approach results in an over-estimate of effect size for two reasons: (1) LOD scores, which are maximized over numerous point-wise tests, are correlated with estimated effect size and (2) small effect QTLs are often undetected in underpowered experiments, yielding inflated effect sizes for detected QTLs (the Beavis effect). When it is impractical to maintain or generate large population sizes, an alternative is to use two populations, one for initial detection and localization and a second for a locus-matched estimate of effect size, not conditioned on significance. Recombinant inbred (RI) panels are eminently suitable for this approach, as each strain genotype can be sampled repeatedly. We present mapping results from the LXS RI panel for two behavioral phenotypes relating to ethanol response: low-dose ethanol activation and loss of righting following high-dose injection. Both the phenotypes were measured in two or three independent cohorts, which were then used to re-estimate effect size. Many small-effect QTLs replicated using this approach, but in all cases, effect size, in the replicate cohorts, was reduced from the initial estimate, often substantially. Such a reduction will have important consequences for power analyses in which sample sizes are determined for subsequent confirmation studies.
Subject(s)
Alcoholism/genetics , Behavior, Animal/physiology , Chromosome Mapping/methods , Quantitative Trait Loci/genetics , Quantitative Trait, Heritable , Animals , Behavior, Animal/drug effects , Bias , Cohort Studies , Computer Simulation , Crosses, Genetic , Disease Models, Animal , Ethanol , Female , Genetics, Population , Lod Score , Male , Mice , Mice, Inbred Strains , Models, Genetic , Phenotype , Reflex/drug effects , Reflex/genetics , Reproducibility of Results , Research Design , Sample SizeABSTRACT
The Colorado IDDM Registry identifies newly diagnosed cases of insulin-dependent diabetes mellitus (IDDM) throughout the state. Hispanics in Colorado are a racial mixture of American Indian and White populations. Because American Indians have a low risk of IDDM, and differing frequencies of HLA antigens and haplotypes are reported for Hispanics and non-Hispanics, we compared incidence rates and disease characteristics. Eligible participants were less than 18 yr of age and Colorado residents at time of diagnosis, diagnosed between 1 January 1978 and 31 December 1983, and on insulin within 2 wk of diagnosis. Subjects were reported by their physicians, and statewide validation of reporting was conducted through review of hospital discharge indexes. Incidence rates for Hispanics (n = 76) were significantly lower than those for non-Hispanics (n = 628), although 95% confidence intervals overlapped for children aged 10-17 yr. Age-adjusted rates were significantly lower in Hispanic than non-Hispanic males, whereas age-adjusted rates for females did not differ. The cumulative risk of IDDM was less for Hispanic males aged 0-17 yr than for non-Hispanic males (P less than .001); cumulative risk among females was males (P less than .001); cumulative risk among females was not different (P = .10). Clinical onset characteristics and medical care at diagnosis were similar. After diagnosis, hospitalizations per 100 person-yr appeared higher in Hispanics, but ketoacidosis and insulin reactions per 100 person-yr were similar. Difference in rate of hospitalizations may have been due to lower response rates among older non-Hispanics.(ABSTRACT TRUNCATED AT 250 WORDS)
