FIMO negative scores

[Rocky Parida]

Hi All

I am using FIMO to find significant occurrences of motif in a set of sequences. I am calling an occurrence of a motif significant if q-value <0.05. I am seeing negative scores for some motif occurrences that also have q-value <0.05.
Is there a lower limit on what log-likelihood ratio score is valid for FIMO analysis?
I appreciate any comments.
Thanking you
Rocky

[CharlesEGrant]

The range of possible log-likelihood scores varies with the length of the motif, the background model, and the information content of the motif. For a given motif and background model you can use the log likelihood score to compare the relative quality of two matches, but a negative score in itself doesn't indicate a "poor" match. It may be that the range of possible log likelihood scores for a motif runs  from -1e-4 to -1e-10, and achieving a score as high as -1e-10 by chance is extremely unlikely. 

For each motif FIMO estimates the distribution of possible scores, and uses the computed distribution to compute measures of statistical significance for each observed score. The p-value or q-value is what you should threshold on, not the raw score.

[Rocky Parida]

Hi Charles.
Appreciate this useful information and it helps.
Does meme has a tool that can draw feature maps like RSAT does, given a location of a motif on a sequence it draws a map as shown here?
Please let me know.
Thanking you
Rocky

[CharlesEGrant]

MEME provides a similar diagram for the motifs it discovers, but FIMO does not. The GFF output of FIMO can be uploaded as a custom track to the UCSC or Ensembl Genome browsers though.

[Rocky Parida]

Thank Charles. Appreciate it.