Methodological development is an essential part of DCMB research. Oftentimes, methodological development is driven by practical research questions, but the innovative and impactful tools and methods take on lives of their own and have broad impact in the field of computational biology. The vast majority of DCMB faculty members ground their research on both practical questions and methodological development, making significant contributions on theories, generalizable methods and their applications to biomedical sciences.
In the area of genomics, our methodological development includes large-scale association analysis, meta-analysis and imputation (Willer, Abecasis), analysis of complex regions of the genome that are not easily resolved through modern sequencing approaches and the integration of multiple types of genomic data (Mills, Boyle, Rajapakse), algorithms for integrating multiple types of genomic data (Guan, Rajapakse), and pattern recognition and the analysis of evolutionary history in genomics data (Li). Our research also includes developing computational methods for processing and analyzing complex proteomic datasets (Nesvizhskii). On tools and methods with direct clinical implications, we develop signal/image processing and machine learning methods to create computer-assisted clinical decision support systems that improve patient care and reduce the costs of healthcare (Najarian), and mathematical models of human behavior that are experimentally verifiable (Forger).
Some of our faculty members develop statistical theory, methods and software tools that have wide applications to biomedical research, and these include the analysis of large-scale, high-dimensional, heterogeneous and complex biomedical data (Guan); longitudinal and survival analysis, cure models, missing data, and other statistical methods for bioinformatics (Taylor); mathematical, computational and statistical methods for the design or selection of optimal procedures and experiments and the extraction of maximum information from biochemical data (Schnell). In the area of machine learning, our faculty develop machine learning methods based on matrix manipulation to complement the more common gradient-descent based methods (Guan), and algorithms for feature extraction from high-dimensional data, sparse learning, multi-task learning, transfer learning, active learning, multi-label classification, and matrix completion (Ye).