The identification of DNA binding sites has been a challenge since the
early days of computational biology, and its importance has been
increasing with the development of new experimental techniques and the
ensuing flood of large-scale genomics and epigenomics data yielding
approximate regions of binding. Many binding sites have a pronounced
positional preference in their target regions, which makes them hard
to find as this preference is typically unknown, and many of them are
weak and cannot be found from target regions alone but only by
comparison with carefully selected control sets. Several de-novo motif
discovery programs have been developed that can either learn
positional preferences from target regions or differentially abundant
motifs in target versus control regions, but the combination of both
ideas has been neglected.

Here, we introduce Dispom, a de-novo motif discovery program for
learning differentially abundant motifs and their positional
preferences simultaneously. Dispom outperforms existing programs based
on benchmark data and succeeded in detecting a novel auxin-responsive
element (ARE) substantially more auxin-specific than the canonical
ARE. Since its publication, we have endowed Dispom with more complex
motif models and extended it to handle weighted input data such as
ChIP-seq or BS-seq data. We have been applying Dispom to in-house and
publicly available data of different transcription factors and
insulators in yeasts, plants, and mammals as well as to
protein-binding microarrays, where it turned out to be one of the
top-scoring approaches in the corresponding DREAM challenge.