|
Abstract:
|
Tissue samples obtained from clinical practices are usually mixtures of different cell types. The high-throughput data obtained from these samples are thus mixed signals. The cell mixture brings complications to data analysis, and will lead to biased results if not properly accounted for. In this talk, I will present some of our recent works on methods and strategies for analyzing high-throughput data from heterogeneous samples, including cell type-specific differential analysis and improved reference-free signal deconvolution. I will first introduce a novel method to model the high-throughput data from heterogeneous samples, and to detect different signals. Our method allows flexible statistical inference for detecting a variety of cell type-specific changes. Based on this model, we further develop an algorithm to improve the reference-free cell composition estimation through better feature selection. Extensive simulation and application to real datasets demonstrate the favorable performance of our proposed method compared with existing ones serving similar purpose.
|
