MEME's motifs have an E-vaue too high

[Laura Pineda]

Hi, I'm currently learning how to use MEME-Suite (web tool) to find a specific known motif in the promoter region of 5 putative bacterial genes of interest, but the results obtained with just the five promtoer sequences have an E-value too high. What can I do to improve these results? Can I use as input some other reported sequences that have the motif I'm looking for without introducing any bias?

[cegrant]

Could you attach a copy of your HTML output file? That would help us give better advice.

[Laura Pineda]

I used the option ZOOPS, with a background model of first order, and I limited the search to palindromes only, since I'm looking specifically for a 20 bp sequence with dyad symmetry. The input sequences are the promoter regions of each gene (200 pb upstream of the start codon).

[cegrant]

I'd suggest a few things. First, since you are looking for a known promoter you might consider using FIMO rather than MEME.  MEME is a de novo motif finder, meaning its job is to discover potentially unknown motifs. FIMO on the other hand is designed to look for matches to known motifs. To use FIMO you'll need to supply the Position Weight Matrix (PWM) for the motif in the MEME format. The command line version of the MEME Suite provides a collection of script for translating other popular motif formats into MEME format. Alternatively, you can just select "Type in motifs" under  "Input the motifs" and type in sequences or regular expressions describing the motif. Click on the red question mark icon next to the selection box for more details.

If for some reason you need to stick with MEME, and you know that each of your five sequences contains an instance of the promoter, you should set the minimum number of allowed motifs sites to 5. This setting is found under the "Advanced options" section, labeled "How many sites must each motif have?".

Finally, how wide is your promoter. It looks like you accepted the default value of  6-50. Since you know your promoter, I'd suggest setting the allowed size range to more closely match what you expect. This setting is also under the "Advanced options" section, labeled "How wide can motifs be?". Note that MEME is weak at finding motifs less than 8bp side. If you think your motif is smaller than that, you might try DREME instead. Like MEME, DREME is a de novo motif finder, but it's stronger at finding motifs less than 8bp wide.