In recent years, Protein Binding Microarray (PBM) has been developed to measure the binding preference of a protein to a set of DNA sequences in vitro. The PBM data resolution is unprecedentedly high, comparing with the other traditional techniques. Researchers have started using the technique for a large number of DNA-binding proteins. A large amount of PBM data has been being accumulated and deposited to the UniProbe database. To analyze the PBM data, discovering motif models from them are essential.


To discover DNA motifs on PBM data, a computational pipeline using Hidden Markov Model (HMM) has been proposed and named, kmerHMM. The method has been compared with the state-of-the-arts methods on well-studied datasets. The results demonstrated the effectiveness of the proposed approach. In addition, belief propagations have been applied to reveal its multimodal motif discovery ability validated by wet-lab experiments.

© 2013 Ka-Chun Wong | Template design by Andreas Viklund