A Brief Description
We recently developed a deep learning method that enabled us to identify 88 nucleosome positioning features. This website takes a nucleosome occupancy profile as input and, after smoothing the profile, centers a 20-base window at each local maximum, provided that these maxima do not overlap with other maxima located within 160 bases of each other. Each position within the window is considered a potential dyad position and is scored based on the 88 identified features. The genomic coordinate with the highest score is selected as the dyad position. In addition, the website provides a classifier for nucleosomal and non-nucleosomal DNA and generates a nucleosome positioning score accordingly. NUCLEONE is freely accessible to both academic and commercial users.

This figure illustrates how Nucleone identifies nucleosome positions from a nucleosome occupancy profile, which serves as its input. The x- and y-axes represent genomic coordinates and the dyad count at each genomic coordinate, respectively. The nucleosome occupancy profile is then smoothed using a triweight kernel function, as described in the main text (step A). Next, a 50-base window is centered on the maximum point of the smoothed dyad count profile. Each genomic coordinate within the window is then scored using the nucleosome positioning sequence features specified by our deep learning approach (step B). The genomic coordinate with the highest score is identified as the dyad position (step C). Subsequently, genomic coordinates within 73 bases of the identified dyad position are excluded, and steps A to C are repeated until all dyad positions have been identified.