Finished Genome of the Fungal Wheat Pathogen Mycosphaerella graminicola Reveals Dispensome Structure, Chromosome Plasticity, and Stealth Pathogenesis
Figure 4
Principal Component Analysis of: S, observed genes on the dispensome; O, observed samples of genes on the core chromosomes before mutation; and x, samples of genes from the core chromosomes after mutation.
Mutation was simulated using observed frequencies of all mutations in families of transposable elements with ten or more copies, and included mutations from RIP and other processes. Mutating the samples of genes from the core chromosomes always made them more similar to the observed value for the dispensome but only rarely included the dispensome value (see panel C). This occurred primarily with codon preference and GC content by amino acid, which are the quantities that are least subject to natural selection for protein function. A, amino acid frequency using the values for the aligned sequence with the highest GC content to build the table of mutation frequencies; B, codon preference using the consensus of the aligned sequences to make the table of mutation frequencies covering only the 5′ portion of each gene; C, codon preference using the values for the aligned sequence with the highest GC content to build the table of mutation frequencies covering only the 5′ portion of each gene; D, codon usage using the values for the aligned sequence with the highest GC content to build the table of mutation frequencies but with all mutation frequencies cut in half; E, codon usage using the values for the aligned sequence with the highest GC content to build the table of mutation frequencies; and F, GC skew using the consensus of the aligned sequences to make the table of mutation frequencies. The first principal component always separated out the pre- and post-mutated chromosome samples. The locations of the observed values for the dispensome (S) are circled.