Transposable element subfamily annotation has a reproducibility problem

摘要

Transposable elements (TEs) are usually annotated within a genome using a tool, such as RepeatMasker [1], that compares a genome to a library of known TEs, such as Repbase [2]. In such a library, TE remnants are classified into families and subfamilies. Subfamilies are in some cases included in a TE library in order to increase annotation coverage, but they more often represent the history of replication and divergence of a family. This history can be complex, with numerous replication bursts leading to clusters of related TEs [3, 4]. Standard practice is to reconstruct and define such subfamilies based on shared diagnostic sequence variation [5, 6] within such bursts. Because annotation with these subfamilies is believed to give some indication of a sequence’s historical context, it is important that such annotation be reproducible.