The preferred nucleotide contexts of the AID/APOBEC cytidine deaminases have differential effects when mutating retrotransposon and virus sequences compared to host genes

Figures

• 

  Fig 1. Description of the model used to calculate susceptibility of a sequence to a particular motif. A) The example sequence has 4 AGC motifs, indicated in purple (with two motifs occurring in the reverse orientation as GCT). B) If every AGC and GCT mutation were mutated to AGT and ACT (as indicated in orange) consistent with AID deamination, three out of four of these motifs cause nonsynonymous mutations changing the corresponding amino acid sequence. C) Susceptibility to the AGC motif is calculated by comparing the percentage of random sequences (maintaining protein sequence) with a lower nonsynonymous mutation fraction. D) Since only one random sequence out of four has a lower nonsynonymous mutation fraction than the baseline, this sequence’s percentile, or susceptibility is 0.25.

• 

  Fig 2. Computing average susceptibilities for each genome to a given motif. A) The susceptibility scores in the EBV genome are calculated for each gene together with the average B) Distribution of percentiles in the EBV genome and C) in the human housekeeping gene set, to the AID motif AGC. The average, shown in a bold colored line, is low (close to 0) in a very resistant genome.

• 

  Fig 3. Gene sets cluster by hypothetical NNC motif susceptibility. The clustering suggests that the gene sets we examined segregate clearly into AID/APOBEC-resistant (boxed in red), and AID/APOBEC-vulnerable clusters (boxed in blue). Many of the known hotspots for AID and APOBEC also cluster together (boxed in orange).

• 

  Fig 4. Low GC content is correlated to susceptibility to some NNC motifs. A) Boxplots show the distribution of GC content of APOBEC-resistant (left) and APOBEC-vulnerable (right) gene sets. B) Plot showing the correlation coefficient (Y-axis) between our gene sets’ susceptibility to a particular motif on the bottom axis to its GC content. C) The correlation of a gene set’s mean susceptibility to the hotspot TTC, a GC-rich hotspot, to its GC content (red = resistant, green = vulnerable based on our clustering observations). D) Correlation of a gene set’s GC content to mean susceptibility for the hotspot CCC. E) Correlation between GC content of all gene sets and a measure of susceptibility based on the absolute number of nonsynonymous mutations at motifs. Susceptibility in this case is negatively correlated with motifs beginning with A and T, and more positively correlated with motifs beginning with G and C.

• 

  Fig 5. The naturally observed cytidine deaminase hotspots better distinguish vulnerable and resistant gene sets than hypothetical NNC motifs. A) Boxplot distribution of differences in average susceptibilities between APOBEC-vulnerable and APOBEC-resistant gene sets. Canonical AID hotspots are in red, and APOBEC3A/B hotspots are in blue. The degenerate motifs WRC and NTC, are labeled in white. The top cluster of motifs (boxed in orange) are enriched in these motifs. B, C) Violin plots of separate gene families’ susceptiblities, clustered (left family in red, right in green), with protruding black lines representing the mean for that gene set. Motifs are sorted by the greatest difference in mean susceptibilities, with many of the motifs showing the greatest difference having the motif NTC or WWC (W = A/T).