MEME stops without errors after finding the first motif

[Joel Wallenius]

The terminal input is:

meme inp001.fna -dna -mod oops -nmotifs 5 -evt 0.01 -w 8 -maxsize 20000000 -p 4

The terminal output is:

w set, setting max and min widths to 8
Initializing the motif probability tables for 2 to 236 sites...
nsites = 236
Done initializing.
SEEDS: highwater mark: seq 235 pos 1133
BALANCE: samples 236 residues 225584 nodes 1 residues/node 225584

seqs=   236, min= 502, max= 7732, total=   225584

motif=1
SEED WIDTHS: 8
em: w=   8, psites= 236, iter=  40


And then all processing is terminated. Doesn't matter if I put nmotifs 5 or 9 or 666. I don't know what is going on. No error messages anywhere.

Very grateful for any help!

[Joel Wallenius]

Uhm it seems to have been a problem with bash loops and subshells... alternatively it could be the -evt flag messing things up.

I seem to have fixed it now, either way. Sorry for taking up post space!

[CharlesEGrant]

Be sure to read the fine print about the -evt flag:

MEME will stop searching for motifs if the last motif found has an E-value > ev.

On top of that, MEME is not guaranteed to return motifs in strictly increasing order of E-value. This means that the -evt option is rarely useful.

[Joel Wallenius]

Thank you very much Charles!

Up to this point, MEME has found the motifs in order of significance.

Is it generally safe to say that a motif with E-value > 1 is random noise? Or must I calibrate that somehow?

I would very much like to reduce the computation time. Right now I'm finding motifs, 80 % of which have E-values above 1e500...

[CharlesEGrant]

You can treat the E-value a p-value controlled for the size of the input sequence data. The traditional significance threshold for p-values is 0.01 or 0.05. This is only a convention though. See https://en.wikipedia.org/wiki/E-value#History