|
Abstract:
|
Efficient detection of transcription factor binding sites is a well known problem in computational genomics. Due to the poor predictive ability of motif finding algorithms, researchers, recently, have been taking advantage of secondary structures, like nucleosomes, for improving predictions. Nucleosomes are compact chromatin structures which wrap the DNA sequence around at specific positions, and prevent transcription factors to bind at those places. Our goal was to predict the nucleosomal free regions , based on gene signal, through a hidden Markov model. However, the gene signal data obtained from recent procedures like FAIRE often has missing data. This feature was accounted for by a continuous time version of HMM. Also, we incorporated DNA sequence features, in a novel way, into our model for improved prediction.
|
