Development of methods to analyze the relationship between chromatin and splicing
The search of sequence prerequisites for nucleosome positioning has been a long-standing problem at the intersection of chromatin structure and gene regulation. With the use of structural information directly related to the primary DNA sequence we propose a new property of natural nucleosome forming sequences, which is inherently related to their intrinsic curvature and its symmetry. A measure of this property has been introduced and a corresponding method validated against novel high quality datasets of human nucleosomes, obtained through close collaboration with CRG's Chromatin and Gene Expression Group (Miguel Beato). Based on this symmetrical curvature property, we have implemented a computational ab initio method for nucleosome positioning prediction and have shown this method to be more efficient than previously published ab initio methods aiming at the same goal. The aforementioned method is already being applied in a number of analyses related to nucleosomal patterns of regions of interest in various organisms. The results have contributed to the formulation of the “key nucleosome” concept. In a study carried out by our group on nucleosome positioning (Tilgner et al., 2009) we employed a computational tool developed in-house (SymCurv) which consists of a computational ab initio method for nucleosome positioning prediction based on the structural property of natural nucleosome-forming sequences to be symmetrically curved around a local minimum of curvature. This tool was first used to help uncover structural constraints in consistent nucleosome positions in the genome of S. cerevisiae (Nikolau et al., 2010)
Furthermore, during 2011 we have used a statistical framework similar to that used to model transcription activity to help us develop predictive models to explore the relationship between levels of histone modifications in the vicinity of exons and the corresponding exon inclusion levels. The contribution of chromatin modifications to splicing is not expected to be dominant, and as a consequence, the predictive power of the exon inclusion models is smaller than that of the models of transcriptional activity. Nevertheless, a number of chromatin marks are identified consistently across cell lines as having a significant association with exon inclusion levels. Some with a strong positive association with exon inclusion, others with a clear negative one. Importantly, the models inferred in a particular cell type are, in general, quite accurate in the other cell lines, suggesting that the relationships between histone modifications and alternative splicing uncovered here appear to be general. Results derived from this work are being compiled to be submitted as companion articles to the main ENCODE paper to the journal Genome Research (“Deep sequencing of RNA from distinct subcellular fractions shows that splicing in the human genome occurs predominantly during transcription” and “Deep sequencing of RNA from distinct subcellular fractions shows that splicing in the human genome occurs predominantly during transcription”) in early 2012.