TF binding motif occurrence-based features (Group II)

Previous studies have shown that DNA regulatory regions are occupied by many TFs^2,22. Generally, the binding sites of TFs can be characterized by Position Weight Matrix (PWM). In this work, a total of 1,422 PWMs were collected from Cistrome⁴⁵ and the TRANSFAC track in the UCSC Genome Browser⁴⁶. For a given sequence, the ability of a TF to bind DNA was represented by the transcription factor affinity prediction (TRAP) score⁴⁷, which was calculated according to the transcription factor motif PWM by the TRAP⁴⁷ program. The parameters of the TRAP program used default values. Because one PWM generates a TRAP score, there were 1,422 TRAP scores for a given sequence. This feature group was designated “TF binding motif occurrence” and contained 1,422 features.