
Advanced options
In the course of building and using a motif-based hidden Markov
model, Meta-MEME requires that various parameters be set. By
default, Meta-MEME employs reasonable values for these parameters.
However, when submitting a job to the server, you may choose to
override the default settings. The following list explains each
parameter and gives its default value.
Model building
- Model topology
By default, Meta-MEME builds models that have a linear topology, in
which the motifs are arranged like beads on a string. If, however,
you believe that the family you are interested in contains repeated
or shuffled elements, you may wish to build a model in which every
motif is connected to every other motif. Note that completely
connected models cannot be used to create multiple alignments.
- Motifs
By default, Meta-MEME includes in the hidden Markov model all the
motifs that appear in more than half of the sequences in the
training set. You may choose instead to include more or fewer of
the given motifs.
- Spacer models
Meta-MEME models the non-motif, spacer regions of the given family
using HMM states with self-loops on them. Because a single such
state yields an exponential distribution of spacer lengths,
Meta-MEME can use several such states in succession to give a more
bell-shaped distribution of spacer lengths. Note that these
multi-state spacer models are only useful in conjunction with total
probability scoring.
Homology detection
- Scoring
By default, when Meta-MEME uses a model to search a sequence
database for homologs, the sequences are scored using the Viterbi
algorithm. This algorithm computes the most likely path through the
model, given the sequence. Alternatively, you can ask Meta-MEME to
compute the total probability of the sequence with respect to all
paths through the model.
- Score threshold
Meta-MEME reports homology log-odds scores in bits. Typically,
Meta-MEME only reports scores for sequences that score above zero.
You can set the score reporting threshold to some other value. Note
that, regardless of the threshold you select, the Meta-MEME server
will report at least 10 scores and at most 1000 scores.
- Minimum number of sequences
This parameter determines the minimum number of homology scores
reported by Meta-MEME. The default value is 10.
- Maximum number of sequences
This parameter determines the maximum number of homology scores
reported by Meta-MEME. The default value is 100.
Return to the Meta-MEME home page.
Please send comments and questions to Charles E. Grant at
.