Abstract:
|
With the rapid advancement in science and technology, the proliferation of large-scale data with various types and formats has been growing at an unprecedented rate. Each of these distinct data types provides a different but complementary view of the corresponding scientific problem, which makes systematic integration an essential component in many fields of science. Despite the need for powerful and advanced integrative analysis strategies, it still lacks the coherent statistical framework for data integration. In this talk, I will introduce a novel and unified framework for data integration. In the proposed framework, we observe multiple measurements that measure the same underlying quantity through some linear or nonlinear ways. We seek to find a fused measurement, which is a scalar representation of those observed measurements and highly correlated with the underlying quantity. We also established the asymptotic property of the fused measurement and integrated multiple histone modifications and DNA methylation levels to jointly characterize epigenetic activeness.
|