Quantitative trait locus analysis using recombinant inbred intercrosses: theoretical and empirical considerations

Abstract

We describe a new approach, called recombinant inbred intercross (RIX) mapping, that extends the power of recombinant inbred (RI) lines to provide sensitive detection of quantitative trait loci (QTL) responsible for complex genetic and nongenetic interactions. RIXs are generated by producing F1 hybrids between all or a subset of parental RI lines. By dramatically extending the number of unique, reproducible genomes, RIXs share some of the best properties of both the parental RI and F2 mapping panels. These attributes make the RIX method ideally suited for experiments requiring analysis of multiple parameters, under different environmental conditions and/or temporal sampling. However, since any pair of RIX genomes shares either one or no parental RIs, this cross introduces an unusual population structure requiring special computational approaches for analysis. Herein, we propose an efficient statistical procedure for QTL mapping with RIXs and describe a novel empirical permutation procedure to assess genome-wide significance. This procedure will also be applicable to diallel crosses. Extensive simulations using strain distribution patterns from CXB, AXB/BXA, and BXD mouse RI lines show the theoretical power of the RIX approach and the analysis of CXB RIXs demonstrates the limitations of this procedure when using small RI panels.

Figure 1.—

Production of RIX hybrids. The relationship between the parental strains and the derivative RIs along with the relationships between RIXs is shown.

Figure 2.—

Power comparison of RI and RIX. A QTL with a series of additive (a) and dominant (d) effects was simulated using genotypes from (A) CXB and (B) BXD. Thresholds were determined using 10,000 simulations and 2000 experiments were performed for each level of additive effect. The solid line (2) is the RI power curve and dashed lines are the RIX power curves corresponding to different dominant effects: (1) no dominant effect; (3) d = a/2; (4) ; (5) ; and (6) d = a.

Figure 3.—

Comparison of permuted and empirical thresholds for RIXs. Ten different realizations of the polygenic effect (x-axis) were simulated for RIXs generated from the (A) AXB/BXA and (B) BXD RI sets. The empirical 95th percentile threshold was estimated from the maximal LOD score obtained from 10,000 simulations where data were simulated under the null with every realization of the polygenic effect. For permutation, 10 data sets for each realization of the polygenic effect and their 95th percentile permuted thresholds were calculated. +, permuted thresholds of 1000 simulated data sets under different realizations of the polygenic effect; E, empirical thresholds under different realizations of the polygenic effect.

Figure 4.—

Significance thresholds and permutation distribution of LOD scores for RIXs. Distributions are shown of maximal LOD scores of the data set used to generate Figure 3 for RIXs generated from (A) AXB/BXA and (B) BXD using 5000 permutations of one data set (additive effect = 1.7) that was simulated with one major QTL (results from other simulated data sets show similar patterns). The solid line is the maximal unpermuted LOD score; the dotted line is the 95th percentile of the permuted maximal LOD scores.

Figure 5.—

Significance thresholds and permutation distribution of LOD scores for CXB RIXs. (A) Simulations performed for RIXs generated from CXB as described in Figure 4. (B) Permutation results using body weight to identify the 95th percentile threshold. The solid line is the maximal unpermuted LOD score. The dotted line is the 95th percentile. Solid and dotted lines overlap in B.

Figure 6.—

Localization of body weight QTL. Results for are shown (B) RIs and (C) RIXs generated from CXB and (D) F₂'s from the same parental strains compared to (A) locations of known body weight QTL. Body weight data were adjusted for age, sex, and the interaction between age and sex. Lines in A are regions known to harbor body weight QTL detected in crosses from many different strains. Lines in B–D represent LOD scores. Dotted lines distinguish individual chromosomes. The significance thresholds determined from permutations are not marked since they are higher than any of the resulting curves.

Figure 7.—

Localization of brain weight QTL. Results for (A) RIs and (B) RIXs generated from CXB and (C) F₂'s from the same parental strains are shown. Brain weight data were adjusted for age, body weight, sex, and the interactions between age and sex and sex and body weight. Lines represent LOD scores. Dotted lines distinguish individual chromosomes. The significance thresholds determined from permutations are not marked since they are higher than any of the resulting curves.

Figure 8.—

Distribution of phenotypic variance. Within-strain variance for body and brain weights from RIs and RIXs generated using the CXB phenotypic data. Data were adjusted as described in Figures 6 and 7. Plots represent the range of maximal standard deviations within each representative set while shaded boxes show mid-50th percentiles and boldface lines show the means.