Fig 1.
Overview of the retroduplication calling pipeline.
A—A simplified flow chart of our calling pipeline. B—A schematic diagram of our strategies. We first align unmapped reads to exon junction libraries and use decoy libraries to control the false discovery rate (FDR). Then, we collect discordant paired-end reads, and cluster the reads that are mapped distal to the parent genes. Clustered distal reads indicate retroduplication insertion site.
Fig 2.
Common retroduplication frequency spectrum and phylogenetic tree.
A—Frequency spectrum of 29 retroduplication events that are detected in more than 10 populations. Hierarchical clustering. B—PCA biplot of the populations based on frequencies of these 29 retroduplication events. Different colors indicate five superpopulations, i.e. AFR (African), AMR (Ad Mixed American), EAS (East Asian), EUR (European), and SAS (South Asian). Arrows represent loadings of parent genes. Ad Mixed Americans are marked with ‘*’. C—Consensus phylogenetic tree built based on novel retroduplications from all 26 populations enrolled in the 1000 Genome Project Phase 3. Bootstrap probability (BP) value is computed from ordinary bootstrap resampling. It is the frequency of the cluster appearing in bootstrap replicates. Approximately unbiased (AU) probability value is calculated from multiscale bootstrap resampling [33,34]. AU is less biased than BP. Bootstrap resampling was performed 1,000 times for generating the trees that are summarized in the consensus tree. Manhattan distance and average linkage was used in hierarchical clustering.
Fig 3.
Overlap between retroduplication insertion sites and genomic features/functional elements.
A—Aggregation plot around insertion sites with strongly positioned nucleosomes. B—Association between discordant read clusters that only have support on one side and L1 element subfamilies. Fold change and empirical p-values were obtained from permutations tests. *** indicates adjusted p-value < 0.001. C—Overlap between genomic elements and retroduplication insertion sites. The enrichment of overlap is expressed as log2 fold change of the observed overlap statistic versus the mean of its null distribution. Positive (negative) log2 fold change indicates enriched (depleted) genomic element-insertion overlap, compared to random background. * indicates empirical p-value ≤ 0.002.