A Unified Approach to Signal Detection
Two major classes of methods for finding regulatory signals are homologous and non-homologous. Homologous will find signals by their modified or lower rate of evolution relative to a background sequence. Non-homologous will find signals because they occur with a higher frequency than in a background sequence. These two approaches have been combined before (Wang and Stormo, 2003; Zhou and Wong, 2007). Combining the approaches should have several advantages. Homologous detection is increasingly powerful due to the increase in know genomes. Non-homologous analysis is powerful in making functional statements since it can compare genes with different properties or that are co-regulated. This project proposes to unify the two approaches within the framework of statistical alignment.