Additional File 1.
Formulation of TFBSs prediction problem. TFBSs prediction problem can be formulated as a function to map a feature matrix (the above matrix in the figure) to an annotation (the below row vector). In the feature matrix, every row corresponds to one features and every column corresponds to one 200 bp bin in a genome. Feature types contain one real value feature (PWM) and multiple binary features (such as "is the bin within a promoter region" and "is it within the peak of a histone marker"). Note that "TSS" stands for transcription start site proximity.
Format: PNG Size: 47KB Download file
He et al. BMC Genomics 2012 13(Suppl 8):S18 doi:10.1186/1471-2164-13-S8-S18