Does `find-repeats` detect only homopolymeric repeats or all repetitive elements?

[AchiadChenzi]

Description

I'm using REDItools find-repeats command to generate a BED file of repetitive regions for filtering RNA editing sites. However, I've observed that the output appears to contain only homopolymeric repeats (consecutive identical nucleotides like AAAAA, CCCCC, TTTTT, GGGGG) and not other types of repetitive elements (such as tandem repeats, microsatellites, or transposable elements).

Command Used

python -m reditools find-repeats --min-length 5 --output repeats.min5.bed reference.fa

Observed Output

The generated BED file (repeats.min5.bed) has the following format:
1 1023 1029 6 C
1 1189 1194 5 C
1 2241 2248 7 T
1 3014 3021 7 A
1 3453 3460 7 A

Column 1: Chromosome
Column 2: Start position (0-based)
Column 3: End position
Column 4: Length of repeat
Column 5: Nucleotide type (A, C, G, or T)

All entries show single nucleotide types (A, C, G, T), indicating these are homopolymeric regions only.

Question

Is this the intended behavior of find-repeats? Specifically:

1. Is find-repeats designed to detect only homopolymeric repeats (consecutive identical nucleotides), or should it also detect other repetitive elements like:

   • Tandem repeats (e.g., ATATAT, GCGCGC)
   • Microsatellites
   • Transposable elements
   • Other repetitive motifs

2. If it's intended to detect only homopolymers, is there a different REDItools command or parameter to detect broader repetitive elements?

3. If it should detect broader repeats, could this be a bug or limitation in the current implementation?

Additional Context

I notice that REDItools also has an --omopolymeric-span parameter in the analyze command that filters sites within homopolymer regions. This suggests that homopolymer filtering is important, but I'm wondering if there's a way to also filter broader repetitive elements using find-repeats or another method.

Thank you for your clarification!

---

Note: If this is expected behavior, I would appreciate any recommendations for tools or methods to detect broader repetitive elements (like RepeatMasker or Tandem Repeats Finder) that could be integrated into the RNA editing analysis pipeline.

[ahanden]

Before answering your questions, a note on repetitive elements and their relation to REDItools. It has been well documented that both sequencing platforms and alignment software do a poor job with repetitive elements. This is the reason many, if not most, pipelines use masked genomes, ignoring these regions entirely. If you do choose to include such regions as input to REDItools, you may find yourself with a high number of false positives and/or inaccurate numbers for AEI.

1. Is find-repeats designed to detect only homopolymeric repeats (consecutive identical nucleotides)

Yes, find-repeats only identifies homopolymeric repeats. In REDItools2, identifying homopolymeric repeats and using them as a filter during the editing detection process was all a part of analyze. For REDI3, we decided to separate detecting the homopolymeric spans to its own function, find-repeats. I do think we should update our documentation to make it clear that this is the only thing detected.

2. If it's intended to detect only homopolymers, is there a different REDItools command or parameter to detect broader repetitive elements?

No, REDItools only has detection for homopolymeric regions.

3. If it should detect broader repeats, could this be a bug or limitation in the current implementation?

This is certainly something we could look into adding. For less-studied genomes, being able to detect other forms of repetitive elements and exclude them from REDItools' output could help filter false positives. I could see implementing this in a future update - but probably nothing for at least the next 4 months.

Until we get around to that, you may want to consider using other tools to identify "problematic" regions that could lead to false positives. In the book "RNA Editing", it's recommended to use RepeatMasker, as you suggested. Rodriguez & Arkhipova, 2023 also has a very comprehensive list of tools for this task.

I notice that REDItools also has an --omopolymeric-span parameter in the analyze command that filters sites within homopolymer regions.

The --omopolymeric-span is left over from the split between analyze and find-repeats. Using the --omopolymeric-span with analyze actually does nothing. This is a known bug that will be fixed in the next release, v3.5.

I'm wondering if there's a way to also filter broader repetitive elements using find-repeats or another method.

You should absolutely filter repetitive elements, as I explained at the start of this comment. How you identify those elements is up to you. If you can ultimately catalogue your repetitive elements in a BED file, you can provide that information to REDItools.

You can tell REDItools to exclude these regions using the -k or --exclude_regions option with a BED file. Note: Someone has pointed out that the exclude reads function is bugged in version 3.4. The issue has been fixed on the dev branch of REDItools3 and will be taken care of for release 3.5