RESUMO
Change in transcription start site (TSS) usage is an important mechanism for the control of transcription process, and has a significant effect on the isoforms being transcribed. One of the goals in the study of TSS is the understanding of how and why their usage differs in different tissues or under different conditions. In light of recent efforts in the mapping of transcription start site landscape using high-throughput sequencing approaches, a quantitative and automated method is needed to process all the data that are being produced. In this work we propose a statistical approach that will classify changes in TSS distribution between different samples into several categories of changes that may have biological significance. Genes selected by the classifiers can then be analyzed together with additional supporting data to determine their biological significance. We use a set of time-course TSS data from mouse dendritic cells stimulated with lipopolysaccharide (LPS) to demonstrate the usefulness of our method.