Extended Data Fig. 6: Effects of mutation rate and selection heterogeneity on polymorphisms and Tajima’s D along genes.

Simulation results from SLiM⁸³ for the first 100 genes of chromosome 1 using a population of 1,000 individuals and 10,000 generations. a–c, Average correlation between 200 permutations of simulated scenarios and observed patterns of variation in natural A. thaliana accessions. a, Parameter choice: Difference between mutation rate (dm) in gene bodies and intergenic space (e.g., 0.5 = 50% reduction in mutation rate) and proportion of mutations that are deleterious in the genic (gds) and intergenic (ids) regions. The parameter combinations shown in d–h are highlighted with red outlines. b, Pearson correlation coefficients (ppc) comparing Tajima’s D values from each simulation to that of observed data in natural A. thaliana accession. c, Pearson correlation coefficients (pcc) comparing number of polymorphisms accumulated in each simulation to that of observed data in wild Arabidopsis accessions. d–k, Examples of polymorphism (red) and Tajima’s D (blue) in relation to gene bodies (TSS, TTS) averaged from 200 permutations of a scenario approximating empirical estimates of mutation rate heterogeneity and selection heterogeneity between gene bodies and intergenic space. Parameters (see a) given for each scenario. Strong purifying selection in gene regions alone (with equal mutation rates between gene bodies and intergenic space), which also reduces levels of polymorphism in gene bodies, causes more negative Tajima’s D values in gene bodies, which is inconsistent with observed data in natural A. thaliana accessions.