Difference between MEME and STREME

[William Holmberg]

I am currently undertaking a project for my final year and was wondering if I could get help to clear something up. I am basically looking at long non-coding RNA in yeast and using point mutations to study their significance. I am using MEME and STREME, but I am not sure what the difference of the tools are. Furhtermore, why I am getting different motifs? 

Thank you in advance for any help 

[cegrant]

MEME and STREME are both de novo motif discovery programs that use statistical method to identify short, similar subsequences that occur more often than would be expected by chance. MEME and STREME use different algorithms to identify these short, similar sequences. STREME is newer, and generally faster, more accurate, and more through than MEME. We'd generally suggest using STREME. MEME is still useful when the number of sequences in the input database is very small (< 50).  In general both MEME and STREME are less effective at identifying RNA protein binding motifs than DNA protein binding motifs, but STREME has a slight edge over MEME. See STREME: Accurate and versatile sequence motif discovery for detailed comparisons.

Note that if you ask MEME to find 50 motifs, it will keep trying to find motifs until it finds all 50, or it runs out of time. Not all of these motifs will be statistically significant! It will generally find the most significant motifs first, but that is not guaranteed. By default STREME will stop when the last three motifs identified exceed the selected p-value threshold. Be sure to look at the MEME E-values and STREME p-values when considering the results.

[BNorthoff]

Thanks for your explanations!

What is the reason to consider E-values of MEME but p-values of STREME?

[cegrant]

My apologies. That was a typo on my part. In both cases you should look at the E-value, as it provides a correction for multiple testing.

Charles