Masking repeats when searching for motifs with MEME

[apredeus]

Hello all,

I've seen quite a few times that people suggest "masking" repetitive regions when performing motif search with MEME. How does one go about it? I know how to get "rmsk" table from UCSC browser; do you overlap these with your bed file obtained when doing the peak calling, and discard intervals that have nonzero overlap?

Also, I was curious about the number of sequences used by MEME-chip. It seems like MEME is limited to 600 but DREME can take more? If anybody has experience with MEME-chip, I'd like to ask few more questions.

Thank you in advance!

-- Alex

[biocomputer]

I've just started using meme for ChIP analysis but I think I can help.

Here's how I get the masked sequences:

1. Have the coordinates of the sites you want to analyze in a bed file.
2. Download a masked genome file, eg. for mouse "chromFaMasked.tar.gz" from http://hgdownload.cse.ucsc.edu/goldenPath/mm10/bigZips/
3. Uncompress the individual chromosome files into a folder, then combine them, this should extract and combine them:

tar zxf ./chromFaMasked.tar.gz -O | cat >> ./mm10Masked.fa

4. Get sequences using bedtools

bedtools getfasta -fi mm10Masked.fa -bed peaks.bed -fo peaks_sequences.fa

You can increase the number of sequences from 600 by specifying the option -nmeme when running meme-chip, see the meme-chip manual page (http://meme.nbcr.net/meme/doc/meme-chip.html).