August 9th 2018: We updated the server to version 1.4.0. (Changelog)
De-novo motif discovery
You can give your job a name to better distinguish between multiple runs.
A fasta file with at least 10 bound nucleotide sequences. Each sequence is assumed to carry at most one motif. If you have long sequences with several motifs, consider chopping them up into several shorter sequences.
When checked, motifs can be both on the plus and minus strand.
The order of the BaMM. A model of order 0 corresponds to a PWM model.
Extends the motif by adding extra positions to the left and right of the initializer motif. If the initializer only consists of the informative core motif, enlarging the motif allows to also learn the flanking regions.
The order for the background model. We recommend a background order of 2 for a realistic model of the genomic input. For very short motifs a background order of 1 or even 0 may be required to find the motif.
A fasta file with sequences that represent the genomic background of the input sequences. If not provided, the background is estimated from the input sequences.
Scan the input sequences for motif occurences. Required for plotting motif localization.
Only motif positions with a p-value smaller than this will be reported as binding positions.
Evaluate the performance of the motif on the input set. Required for plotting performance plots.
Compare motifs with our database.
Motif database used for annotating trained models.
The e-value limit will be used to define a threshold for motif comparisons between the inserted model and the database.
Length of patterns to be searched.
Patterns with a z-score threshold lower than this will not be considered.
Patterns with less than this amount of counts will not be considered.
The scoring function used for the optimization to IUPAC patterns.
When checked, the Expectation Maximization step will be skipped.
Maximum number of seed motifs that will be refined with the BaMM algorithm.