Transposable element annotation of the rice genome

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Annotating repetitive sequences in eukaryotic genomes, particularly transposable elements (TEs), is computationally challenging.
• RepeatMasker is the standard tool for TE identification via pairwise sequence comparison.
• Profile hidden Markov models (HMMs) show promise for TE discovery but are rarely applied to genomic DNA.

Purpose of the Study:

• To evaluate and compare different computational approaches for annotating transposable elements (TEs) in the *Oryza sativa* ssp. *japonica* genome.
• To assess the strengths and weaknesses of RepeatMasker with different search algorithms (cross_match, WU-BLAST) and the HMMER package for TE annotation.

Main Methods:

• Comparative analysis of TE annotation methods including RepeatMasker with cross_match, RepeatMasker with WU-BLAST, and HMMER.
• Utilized Perl scripts for data conversion to GFF format and post-processing.
• Applied HMMER for specific superfamilies (MULEs) and polyphyletic groups (MITEs) with available libraries, while using RepeatMasker for longer elements or those requiring reverse complement searching.

Main Results:

• RepeatMasker with WU-BLAST offered a significant speed improvement over cross_match with minimal loss of sensitivity.
• HMMER was effective for annotating MULE and MITE families but was slow for models >1000 bp and did not inherently search the negative DNA strand.
• Reverse complement models were necessary for HMMER to detect TEs on the negative strand, except for palindromic sequences.

Conclusions:

• WU-BLAST provides a faster alternative to cross_match within RepeatMasker for large-scale TE annotation.
• HMMER is a valuable tool for specific TE families but requires modifications or complementary approaches for comprehensive genomic annotation.
• Integrating profile HMM libraries into RepeatMasker could enhance the detection of degenerated TE copies.