Figure 2

Properties of chimeric transcripts in the D. melanogaster genome. A) Proportions of 59 chimeric TEs in different element classes, compared to those of the 414 non-chimeric TEs found within intronic regions of genes. The proportion of LTR elements among the chimeric TEs is significantly greater than that of TEs found in introns. B) Proportions of 59 chimeric TEs in different length classes, compared to those of the 414 non-chimeric TEs within intronic regions of genes. The "low", "medium" and "high" length classes are defined according to the 33% and 66% length quantiles for the entire set of genomic TEs (748 and 3818 bp, respectively). The chimeric TEs show a significant enrichment for long TE insertions. C) Distribution of chimeric and non-chimeric TEs found within genes, partitioned by different recombination rates. Both distributions are compared against that of the number of genes found in each section of D. melanogaster euchromatin. See Methods for the definitions of "high", "low" and "zero" recombination. Note that the distribution of non-chimeric TEs deviates from the distribution of genes much more significantly than that of the chimeric TEs. In all three panels, the error bars on the numbers of chimeric insertions were obtained by assuming that the intronic proportion is the "true" probability p. Under a normal approximation, we expect the number of chimeric insertions to have mean np and variance np(1-p), where n is the number of chimeric elements. Based on this model, we constructed a 95% confidence interval around the observed number of chimeric elements that corresponds to the error bars in our figure. Error bars on the numbers of intronic insertions in panel C are based on the corresponding proportions of protein-coding genes.