The measured interfacial width consists of a true interfacial width, lt, convoluted with a width, l0.
As an example, the true width of the FruR motif in E. coli is known to be 18bp.
Following [ 26], we define a predicted site as a 'true positive' result if it overlaps the true site by at least a quarter of the true width.
The error between the image and true width is a function of the width of the cantilever tip (Kuznetsov and McPherson 2011).
Retaining the assumption that the motif discovery algorithm discovers the true motif as well as possible at every candidate width, it follows that the model at the true width w∗ will also be removed as a result of it being contained within the model at width w∗ + 1.
It follows that, at candidate widths smaller than the true width (that is, { w min,…,w∗ - 1}), only a portion of the true motif is discovered while at candidate widths larger than the true width (that is, { w∗ + 1,…, w max }), the full motif is discovered, along with a number of background positions.
In transcription factor binding site discovery, the true width of the motif to be discovered is generally not known a priori.
The mean thickness of sandbodies is 2.96 m (n = 118) and mean true width is 112 m (mean W T = 43), with thicker and wider sandbodies in upper levels of the Member.
It follows from the above that an estimator may appear to perform well even if the chosen width does not match the true width [ 12].
At the true width (w∗ = 14bp), the motif is discovered relatively poorly (sSn = 0.27, sPPV = 0.21, AUC = 0.88).
We first note from Table 4 that the width predicted by MCOIN closely matches the true width in almost all cases; the error in the predicted width increases slightly as mean motif conservation is decreased.
