Frequently asked questions
- Why can't the Meta-MEME program 'mhmm' read my MEME output file?
Meta-MEME can only read the HTML version of MEME ouput. Also, Meta-MEME cannot read MEME files that contain both positive and negative motifs. If you don't know what this means, don't worry about it. Only advanced MEME users are likely to produce files containing positive and negative motifs. In the future, Meta-MEME will be capable of processing these motifs.
- Why doesn't Meta-MEME include in the hidden Markov model all of the motifs that I requested?
This is generally only a problem if you have downloaded the Meta-MEME source code and installed it locally.
When building a linear HMM, Meta-MEME uses output from MEME to set the order and spacing of motifs within the model. Meta-MEME therefore requires that the MEME output provided to 'mhmm' contains at least one motif occurence diagram that contains exactly one occurence of each of the selected motifs. Occurences of motifs that were not requested are ignored.
If, for example, you request that the model contain motifs 1, 2 and 3, but the MEME output file contains only these diagrams
30-[1]-23-[3]-1 28-[1]-32 40-[2]-2-[3]-1 2-[1]-[1]-4-[2]-[3]-3then 'mhmm' will be unable to find a representative diagram. The program will then throw out the last requested motif and re-search the diagrams, repeating this procedure until it finds a diagram that works. In this case, there is no diagram containing motifs 1 and 2 either, so the final model would be based upon the first diagram and would contain only motif 1.- All of the scores from my Meta-MEME search are below zero. What does this mean, and how do I fix it?
Occasionally, Meta-MEME has a tendency to improperly scale the log-odds scores that it reports. This is a result of the mismatch between the topology of the foreground and background models. The problem is especially likely when Viterbi, rather than total probability, log-odds scores are computed. You can try computing total probability scores instead, or turn off the training portion of Meta-MEME to create a model that is less finely-tuned to your particular training set.
Return to the Meta-MEME home page.Please send comments and questions to: @METAMEME_CONTACT@.