Bottlenecks -- some ramblings and a bit of data from maize PAGXXII
3 likes3,181 views
1) The document discusses how population bottlenecks during domestication can reduce genetic diversity and influence patterns of variation across the genome. 2) Bottlenecks are shown to increase the number of rare variants, affect site frequency spectra, and influence the effects of purifying selection. They can also mimic signals of positive selection. 3) Studies of maize domestication estimate a weaker bottleneck than previously thought, with rapid recent population growth increasing diversity from resequencing data. However, selection still reduces diversity more in genic regions relative to intergenic regions.
Bottlenecks -- some ramblings and a bit of data from maize PAGXXII
- 1. Thinking about domestication bottlenecks Jeffrey Ross-Ibarra www.rilab.org @jrossibarra rossibarra January 12, 2014
- 2. Diversity lost in domestication bottlenecks Ross-Ibarra et al. 2007
- 3. Bottleneck effects on the SFS
- 4. 10 15 20 25 0 5 π Bottlenecks can mimic selection 20 40 60 80 100 0 20 40 60 80 100 0 −2 −1 D 1 2 0 kb Bottlenecks affect mean π and D, but also inflate variance
- 5. and then determined the probability that the observed S falls into the 95% confident interval of the simulated distribution. Under both recombination estimates, we rejected the multilocus scenario for tb1 ( p ¼ 0.0002 and p ¼ 0.0003, with 4Nchud87 and 4Nchud01, respectively), ts2 ( p ¼ 0.008 and p ¼ 0.0196), d8 ( p ¼ 0.0021 and p ¼ 0.0032), and zagl1 ( p ¼ 0.00 and p ¼ 0.00), strongly suggesting that the bottleneck model alone does not account for the evolutionary history of these genes in maize. In contrast, none of the eight putatively neutral genes could be differentiated from the multilocus model by this method (data not shown). Duration and strength of bottleneck confounded?1 Can estimate k = NB d Size and duration confounded 1 Tenaillon et al. 2004 MBE Previous studies reported a positive and significant correlation between recombination rate, measured by either 4Nchud87 or Wall’s estimator (Wall 2000), and nucleotide diversity (h) in maize (r ¼ 0.65, P ¼ 0.007), based on 18 putatively neutral loci. However, this same correlation was not significant when recombination was measured either by 4Nchud01 or by a physical measure of recombination (R) (Tenaillon et al. 2002). One of the questions we wanted ask was whether the positive correlation observed in maize was also evident in parviglumis. Among seven neutrally evolving loci for ^ which a 4N chud87 value could be determined (table 3), we ^ found no significant correlation between 4N chud87 and h in parviglumis (r ¼ 20.07, p ¼ 0.56). Similarly, h in parviglumis is correlated with neither R (r ¼ 20.116; p ¼ ^ 0.37) nor 4N chud01 (r ¼ 20.25, p ¼ 0.27). By contrast, the ^ correlation between 4N chud87 and h in this subset of seven loci was still high in maize (r ¼ 0.58), but not significant ( p ¼ 0.32), probably reflecting a lack of power with a small sample. Using simulation, we explored whether the population bottleneck could generate the positive correlation between Downloaded from http://mbe.oxfordjournals.org/ by guest on January 10, 2014 Genetic Diversity and Recombination
- 6. distribution. Under both recombination estimates, we rejected the multilocus scenario for tb1 ( p ¼ 0.0002 and p ¼ 0.0003, with 4Nchud87 and 4Nchud01, respectively), ts2 ( p ¼ 0.008 and p ¼ 0.0196), d8 ( p ¼ 0.0021 and p ¼ 0.0032), and zagl1 ( p ¼ 0.00 and p ¼ 0.00), strongly suggesting that the bottleneck model alone does not account for the evolutionary history of these genes in maize. In contrast, none of the eight putatively neutral genes could be differentiated from the multilocus model by this method (data not shown). Duration and strength of bottleneck confounded?1 NB d Size and duration confounded 1 Tenaillon et al. 2004 MBE count Can estimate k = Previous studies reported a positive and significant correlation between recombination rate, measured by either 4Nchud87 or Wall’s estimator (Wall 2000), and nucleotide diversity (h) in maize (r ¼ 0.65, P ¼ 0.007), based on 18 putatively neutral loci. However, this same correlation was not significant when recombination was measured either by 4Nchud01 or by a physical measure of recombination (R) (Tenaillon et al. 2002). One of the questions we wanted ask was whether the 125 positive correlation observed in maize was also evident in parviglumis. Among seven neutrally evolving loci for 100 ^ which a 4N chud87 value could be determined (table 3), we ^ found no significant correlation between 4N chud87 and h in size 75 parviglumis (r ¼ 20.07, p ¼ 0.56). Similarly, big in h parviglumis is correlated with neither R (r ¼ 20.116; p ¼ small 50 ^ 0.37) nor 4N chud01 (r ¼ 20.25, p ¼ 0.27). By contrast, the ^ correlation between 4N chud87 and h in this subset of seven 25 loci was still high in maize (r ¼ 0.58), but not significant ( p ¼ 0.32), probably reflecting a lack of power with a small 0 sample. 800 900 1000 Using simulation, we explored1100 1200 the population whether 1300 π bottleneck could generate the positive correlation between ^ chud87 and h in maize. For this purpose, we performed 4N 10,000 coalescent simulations under the best conditions Downloaded from http://mbe.oxfordjournals.org/ by guest on January 10, 2014 Genetic Diversity and Recombination
- 7. Duration and strength of bottleneck confounded?1 125 100 100 constant exp 50 size 75 count count model big small 50 25 0 0 800 900 1000 1100 1200 1300 pi 800 900 1000 π 1100 1200 1300
- 8. Duration and strength of bottleneck confounded? 100 count size big small 50 0 0.0 0.3 0.6 0.9 1.2 Tajima's D 125 100 100 small 50 size 75 big count count size big small 50 25 0 0 1000 1200 1400 SNPs unique to domesticate 800 900 1000 π 1100 1200 1300
- 9. Bottleneck effects vary over time
- 10. Evolution: Eyre-Walker et al. Previous estimates of the maize bottleneck Single locus adh1 estimates2 reveal loss of diversity Refinement of FIG. 1. Schematic representation of the coalescent models bottleneck and test for selection3 simulation. NB < 0.01 and 45% Analysis of ≈ 800 loci: See text for details. diversity loss4 NA Eyre-Walker et had smaller S than Smaize. In this way, we estimat al. 1996 PNAS Tenaillon et al. expected value and the 95% lower confidence interva 2004 MBE 4 conditional Wright et al. 2005 Science on d, A, P, and Smaize. The 95% lower conf 2 3
- 11. Genome resequencing: more diversity, exponential growth 20000 count 15000 10000 5000 0 0 1 πMZ πTEO 2 3 HapMap 2 data5 show mean loss of diversity only < 20% PSMC6 analysis of resequenced maize landraces estimates bottleneck NB ≈ 0.2 and recent explosive growth7 NA 5 Hufford et al. 2012 Nat. Gen Li & Durbin 2011 Nature 7 Takuno et al., Unpublished 6
- 12. Selection and demography interact along genome Approximate Bayesian Computation of simple growth model estimates stronger bottleneck in genic regions
- 13. Selection and demography interact along genome Nongenic 15000 10000 count Taxa maize teo 5000 0 −2 0 2 4 Tajima's D Excess of new rare intergenic variants in maize8 8 9 Hufford et al. 2012 Nat. Genetics Stoletzki & Eyre-Walker 2011 MBE
- 14. Selection and demography interact along genome Nongenic Genic 15000 3000 10000 2000 maize teo 5000 Taxa count count Taxa maize teo 1000 0 0 −2 0 Tajima's D 2 4 −2 0 2 Tajima's D Excess of new rare intergenic variants in maize8 No excess rare, 40% fewer unique SNPs in genes Purifying selection slows recovery of diversity 8 9 Hufford et al. 2012 Nat. Genetics Stoletzki & Eyre-Walker 2011 MBE
- 15. Selection and demography interact along genome Nongenic Genic 15000 3000 10000 2000 maize teo 5000 Taxa count count Taxa maize teo 1000 0 0 −2 0 Tajima's D 2 4 −2 0 2 Tajima's D Excess of new rare intergenic variants in maize8 No excess rare, 40% fewer unique SNPs in genes Purifying selection slows recovery of diversity Estimates of DFE9 : new nonsynonymous mutations deleterious 8 9 Hufford et al. 2012 Nat. Genetics Stoletzki & Eyre-Walker 2011 MBE
- 16. Patterns of genetic load vary under different models Figure 3 Hi tion in a gro percentage copies. Segr three discre For each ca rived alleles ber of copi across all th gory divided alleles acros population 3 gories sums the last gen with and w bars denote cates. Popu percentage deleterious c Figure 1 Population growth increases the number of segregating sites, but also the fraction of sites that are lost. (A) S, the number of segregating sites population improves the efficacy of natural selection, and of the whole population (on a log scale); (B) %Slost, the percentage of segregating sites lost from the population in a single simulated generation deleterious sites are more readily eliminated. (Materials and Methods). Both panels present the two simulated demographic scenarios (with growth and with no growth) for each selection model (neutral or deleterious). Results are presented every 10 generations (corresponding to a single simulated generation) during the last 440 generations. Fitness effect of deleterious alleles in Population growth increases both S and %Slost. S is smaller for deleterious than for neutral mutations, while %Slost is population with time in the a growing higher. Trends models without growth are due to the preceding population bottlenecks (Figure S3). and nonsignificant (Figure S7 an effect of population growth on the also visible in the site frequency sp generally, the selection coefficient rived allele copies (wDA) become population grows (Figure 4). At t an allele chosen randomly is 15.8 population that has undergone gr that the average selection coe although to a smaller extent—in (Figure 4). This is because the p growth is also not at equilibrium d lation bottlenecks. The role of the dent in comparison to a model of a of constant size throughout history Despite the accumulation of de in the growing population, we sh the average fitness effect of deriv Growth may lead to larger proportion of deleterious SNPs Growth decreases mean effect size of deleterious SNPs To further investigate the effect of genetic drift and natural selection on the number of segregating sites under population growth, we estimated at each generation the percentage of segregating sites that are not observed in the next generation (%Slost). After a few generations of mutation accumulation, %Slost becomes higher for the model with population growth, both for neutral and for deleterious loci (Figure 1B), implying that population growth increases not 10 only the number of segregating sites, but also the rate at Average fitness effect of a deleterious mutation: To go beyond the burden in the number of deleterious mutations and consider their effects, we compared the distribution of 2008; Coventry et al. 2010; Keinan and Clark 2012; Nelson models with and selection coefficients in the population et al. 2012; Tennessen etwithout growth. We computed the average fitness effect al. 2012) by showing an increase in (selection coefficient) for derived alleles the proportion of singletons (sites with DAC = 1) (Figure that are lost and derived alleles that deleterious loci the S2). The proportion is further elevated at are transmitted to for next generation by both population modelsaveraging S2). fitness effect of each ef(Figure the To investigate the allele weighed by its number of copies (Materials and Methods). As expected, in ficacy of purifying selection in a growingscenarios, lostfree of much more delepopulation sites are both demographic the expected skew in the site than sites that are transmitted (Figure S6). Interestterious frequency spectrum, we consider instead the DAC of lost this phenomenon is more pronounced in a growing ingly, segregating sites. population, of lost sites to the higher We computed the percentage again pointing within eachefficacy of selection Gazave et al. 2013 Genetics 10
- 17. More VA due to del. alleles under growth Number of causal variants A 250 ● ● ● 200 150 ● ● ● ● ● 100 ● 50 BN BN+growth Old growth Number of causal variants B 250 200 150 Exponential growth leads to more rare causal variants11 and these explain a larger proportion of VA ● ● ● Standard GWAS has lower power to detect rare deleterious variants ● 100 ● ● ● ● ● BN+growth Old growth ● 50 BN 11 Lohmueller 2013 arXiv
- 18. Conclusions: thinking more about bottlenecks Bottlenecks affect genome-wide diversity and the SFS May lead to false positive signals of selection Bottlenecks and growth affect fate of deleterious variants and genetic architecture of quantitative traits In maize, genome-wide data suggest a weaker bottleneck and rapid population growth Interplay of selection and demography leads to different patterns in genic and intergenic regions