Powerful decomposition of complex traits in a diploid model

Abstract

Explaining trait differences between individuals is a core and challenging aim of life sciences. Here, we introduce a powerful framework for complete decomposition of trait variation into its underlying genetic causes in diploid model organisms. We sequence and systematically pair the recombinant gametes of two intercrossed natural genomes into an array of diploid hybrids with fully assembled and phased genomes, termed Phased Outbred Lines (POLs). We demonstrate the capacity of this approach by partitioning fitness traits of 6,642 Saccharomyces cerevisiae POLs across many environments, achieving near complete trait heritability and precisely estimating additive (73%), dominance (10%), second (7%) and third (1.7%) order epistasis components. We map quantitative trait loci (QTLs) and find nonadditive QTLs to outnumber (3:1) additive loci, dominant contributions to heterosis to outnumber overdominant, and extensive pleiotropy. The POL framework offers the most complete decomposition of diploid traits to date and can be adapted to most model organisms.

Figure 1. An experimental framework for analysis of diploid traits.

(a) Experimental design. Left panel: Advanced intercrossed lines were constructed by multiple rounds of random mating and sporulation of North American (NA) and West African (WA) genomes. Middle panel: We sequenced 172 of the resulting segregants and paired these to generate an array of 7,310 diploid hybrids (POLs). Right panel: The POLs and their F12 haploid parents were growth phenotyped in nine environments, providing high resolution growth curves. (b) Frequency of homozygotes (red: WA/WA, blue: NA/NA), heterozygotes (purple: NA/WA) and missing genotypes (white, mostly attributed to chr. IX aneuploidies) at each segregating site among the 7,310 POLs. Deviations from 50% heterozygosity are explained by selection (numbers 1, 4–8) against one allele in the F12 haploid parent construction, or by forced heterozygosity at the LYS2 (number 2) and MAT (number 3) loci. (c) Growth rate distributions of POLs (blue), their haploid F12 parents (orange) and the diploid parent estimates (grey, Methods). (d) Correlations (Pearson's r) between the growth rate and mean growth for POLs within environments (lower left to upper right diagonal; orange borders), between growth rates (above diagonal) and mean growth (below diagonal) in pairs of environments. Colour intensity (3-colour scale: dark yellow to white to dark blue) and number indicate the degree of correlation r.

Figure 2. Near complete variance decomposition of diploid traits.

Decomposing the total variance in growth traits across 6,642 diploids into additive (grey, upper panel), dominance (yellow, upper panel), second order epistatic (blue, upper panel) and third order epistatic (green, lower panel) genetic contributions. Black label=growth rate, red label=mean growth. Error bars=s.e.m.

Figure 3. Cost-efficient QTL mapping in yeast POLs.

QTLs were mapped across 6,642 genomes and 18 traits based on additive and nonadditive contributions. QTLs were validated as additive or dominant genetic contributions using Linear Mixed Models (LMM). (a) QTL signal strength (LOD score, y-axis) as a function of genomic position (x-axis), for growth rate on allantoin as sole nitrogen source, using additive (LMM and non-LMM; lower panel) and nonadditive (non-LMM and LMM only capturing dominance; upper panel) models. Red dots indicate significant (FDR, q=10%) QTL calls. White/grey fields indicate chromosome spans. (b) Venn diagram of significant QTLs capturing additive and nonadditive genetic contributions. All 18 phenotypes (growth rate and mean growth over nine environments) were considered, with pleiotropic QTLs counted multiple times. (c) Tukey boxplot showing the fraction of variance explained by additive (purple) and nonadditive (blue) significant QTLs (non-LMM models). (d) Histogram of pleiotropic QTLs. A QTL was counted as shared across environments if peaks were within 10 kb of each other. No QTLs were significant in 4, 5, 6 or 7 environments.

Figure 4. Explaining heterosis by intralocus interactions.

(a) Frequencies of the heterotic POLs (y-axis) as a function of a range of FDR significance cut-off (q) values (x-axis). Line colour=type of heterosis. Red text=FDR q-value chosen for downstream analysis (a,c). (b) Left panel: example of QTLs called as contributing to best parent heterosis by dominance (dark orange) and by overdominance (light orange) respectively. Dominance was called as enrichment of strongest homozygote and overdominance as enrichment of heterozygous state among BPH POLs as compared with all POLs (left panel). Right panel: phenotype (top: allantoin, bottom: galactose) distribution depending on genotype composition at the same QTLs. (c) The frequency of QTLs called as contributing by enrichment of the best homozygote, dominance and overdominance respectively (y-axis) as a function of FDR significance cutoff (q) values (x-axis). The dominance contribution is a subfraction of the contributions from enrichment of the best homozygote. Note: we show the outcomes of a range of FDR cut-off values to illustrate the robustness of conclusions; the cut-offs used for downstream analysis was set beforehand and not influenced by the results. Best parent heterosis (BPH); mid parent heterosis (MPH); worst parent heterosis (WPH).