BGI Mobile Element Annotation

Mobile elements in the genome assembly were identified at both the DNA and protein level. RepeatMasker v3.2.9 (http://www.repeatmasker.org/) was applied to screen the genome assembly for low complexity DNA sequences and interspersed repeated elements using a custom library (a combination of Repbase v16.02 and plant repeat database). RepeatProteinMask (an extension of RepeatMasker) was used to perform RMBlast against the ME protein database to find known repeat sequences at the protein level.

Ab initio prediction program RepeatModeler version 1.0.4 (http://www.repeatmasker.org/RepeatModeler/) was employed to build the de novo repeat library from the assembled genome, refined by removing the contaminated sequences possibly derived from bacterial and redundant duplicated sequences in the library. Using this library as a database, RepeatMasker was implemented to identify and classify homologous repeat elements in the genome. In addition, LTR_FINDER version 1.1.0.5 was used to search the whole genome for the characteristic structure of the full-length long terminal repeat (LTR) retrotransposons. Subsequently, a custom program was used to merge all the predictions and generate a combined repetitive sequence annotation to mask the carrot genome.

ME accounted for 44.9% (190 Mb) of the assembled carrot genome. This value is larger than those observed in other sequenced genomes of similar size, for example, grape (41.4%, for 487 Mb) and melon (20%, for 375 Mb). With 57.4 Mb, the fraction of class II transposable elements in the carrot genome is higher than in most other plant genomes including rice (48 Mb). A large fraction of MEs are of relatively recent origin, with a sequence divergence rate of less than 10%.

Data from this analysis can be viewed in JBrowse here.