We previously showed that DME outperforms other motif discovery algorithms on both synthetic and mammalian data.
Two other motif discovery algorithms, MEME [54] and Mdscan [55], were also attempted for de novo motif discovery; however, unlike the robust Gibbs Motif Sampler, either their performance in detecting the consensus DBE with a set of positive control sequences was poor or the parameter setting was not flexible.
There are two ways of finding shared motifs in sets of sequences: (i) applying ab initio motif discovery algorithms which search for recurring patterns in a set of DNA sequences, or (ii) assessing whether previously characterized motifs present in transcription factor binding site databases are statistically over-represented in the sequences [47].
MEME-ChIP employs two motif discovery algorithms with complementary characteristics.
Furthermore, motif discovery algorithms cannot guarantee to return the best possible result at each candidate width.
Further, we used motif discovery algorithms to identify novel upstream regulatory sequences.
Moreover, we have shown that combining priors is very promising in further extending the power of motif discovery algorithms.
Position-specific priors (PSP) have been used with success to boost EM and Gibbs sampler-based motif discovery algorithms.
Among the different motif discovery algorithms supported by this task manager, we ran oligo-analysis [ 18] and dyad-analysis [ 43].
Different motif discovery algorithms applied to the same set of TF binding sequences usually produce different results [ 1].
We ran all motif discovery algorithms available in the web site (oligo-, position-, local-word- and dyadanalysis).
