Modeling and integrative analysis with applications to DNA replication, cancer, and epigenetics
MetadataShow full item record
Biological organisms have evolved complex epigenetic mechanisms to tailor their gene expression programs to specific needs. These adaptations allow cells, that otherwise have identical genomes, to carry out specialized functions. In this work I develop and use data-integrative techniques to examine the mechanisms and consequences of epigenetic processes. To better understand the changes in DNA methylation landscape that accompany breast cancer molecular subtypes, I integrated DNA methylation and gene expression data from 208 breast cancer samples obtained from a Polish population-based case-control study. Using a weighted correlation network approach, I identified gene co-methylation modules and asked if the genes in these modules are preferentially methylated and silenced in a breast cancer subtype-specific manner. This approach identified two non-overlapping gene co-methylation modules. The first module is silenced in Basal breast cancers, while the second is silenced in Luminal B breast cancers. Gene-set enrichment analysis suggests that epigenetic silencing of these modules interferes with processes that maintain cellular differentiation, and that the methylation status of the Luminal B module is associated with disease prognosis. To uncover the determinants of the temporal order of metazoan genome replication, I used a reductionist model of DNA replication to test the ability of hundreds of epigenetic marks to predict replication timing. My work showed that DNA replication timing can be completely predicted from locations of DNase I hypersensitive sites. I further demonstrated the robust emergent character of DNA replication that could be understood without invoking a complex regulatory mechanism. To determine the underlying cause of cell de-differentiation in osteosarcoma, I examined the relationship between microRNA expression and the bone-cell differentiation program. Focusing on the inhibitory role of miR-23a in bone differentiation, I analyzed the effect of its over-expression in osteosarcoma cells. Extensive computational analysis led me to propose that a major mechanism by which miR-23a exerts its effect is by interfering with expression of GJA1, which encodes a gap junction channel essential for intercellular communication and external stimuli sensing in bone cells. Follow-up experiments indicate that GJA1 is sharply up-regulated during bone cell differentiation and that GJA1 inhibition significantly delays the onset of differentiation. Together, this work uses data integrative techniques to provide new insights into the decisive role of epigenetic processes in cellular differentiation.