The Division of Computational Biology advances basic science, translational science, population health and patient care through a wide variety of specialized research focus areas.
Genetic epidemiology and statistical genetics
Researchers in the fields of statistical genetics and genetic epidemiology develop new methods and improve existing workflows to identify genetic risk factors for complex clinical traits. The result of these efforts is a deeper understanding of disease etiology.
Cutting-edge analytical approaches are combined with large-scale clinical and genomic data to discover genes related to disease and clinical outcomes. The division's studies cover a broad range of common and rare diseases including cancer, neuropsychiatric diseases, cardiovascular disease and others using a variety of study designs to better understand how genetic variants influence disease within and across populations. Through biomedical discovery, these studies help predict who has an increased risk of disease. These studies also facilitate treatment development, leading to better diagnosis and treatment of patients.
The division actively pursues studies aimed at deciphering the contribution of common and rare genetic variants, measured using genotyping and sequencing technologies. To facilitate such studies, division researchers develop novel statistical methods for genetic data analysis, including methods to study pleiotropic genetic effects, polygenic risk prediction, gene-environment interaction analysis, sex chromosome analysis and data integration including mediation analyses to link inherited variation with intermediate risk factors leading to disease.
Functional annotation of genomes improves the understanding of genetic mechanisms leading to disease risk, symptoms or prognosis and is important for interpretation of the multitude of personal genomic variations in relation to various diseases and cancer. Such interpretation is one of the most important challenges for biomedical discovery and individualized medicine. Because most genetic variants occur in noncoding regions of the genome, their effects remain poorly understood.
Modern technologies in genomics allow for high-throughput variant discovery, measuring expression of all genes and determining epigenomic alterations across the entire human genome. Studies are conducted to map gene expression and the associated epigenomic influence across multiple diseases, cancers and tissues at the single-cell level. Recent advances have enabled application of technologies such as assay for transposase-accessible chromatin sequencing (ATAC-seq) and RNA sequencing (RNA-seq) to single cells. Expression and epigenome maps based on single-cell analyses enable the deciphering of functions of noncoding regions in a cell-specific way, which is the foundation for revealing how variants contribute to disease pathogenesis, cancer development and progression.
Pharmacogenomics and personalized medicine
In addition to their utility in investigations of disease risk, statistical genetic and genetic epidemiology approaches can be used to identify variants that impact how individuals respond to particular medications. This study of the genetic contribution to drug response, known as pharmacogenomics, provides insights into mechanisms of drug action, increases understanding of factors that contribute to treatment outcomes and ultimately guides development of personalized treatment strategies.
Bioinformatics methodologies and software are developed to interrogate complex sequencing data and apply systems biology techniques to answer diagnostic and pharmacogenomic questions, thereby helping to push the boundaries of understanding the genetic drivers behind a wide variety of human diseases, including cancer.
Custom treatment plans built around genetic aberrations may impact patients' protein interactions, cellular behavior, organ function and bodily health. Devising such plans requires a deep understanding of sequence analysis, pharmacogenomics and systems biology. Experts in the Division of Computational Biology collaborate with researchers and clinicians to discover genomic variations that require customized treatment plans and to support them in building genetically customized care plans.
Computational and statistical metagenomics
The collection of microorganisms found in the human body (the human microbiome) plays a vital role in health and disease. The human microbiome is being studied with the aid of metagenomic sequencing technologies, using either a targeted or a shotgun approach.
Computational and statistical tools are developed and applied in the analysis of metagenomic sequencing data. These important resources support institution wide disease-focused microbiome research and help clinical investigators understand the role of the human microbiome in disease susceptibility, initiation, progression and response to treatment, and ultimately to integrate microbiome data into individualized medicine.
Cancer genomics and evolution
A new horizon in the study of cancer is how scientists classify, quantify and make sense of changing genomic signatures within and across tumors over time. The spatial and temporal evolution of cancer cells is interrogated through the analysis of single cells. These studies will enable scientists and clinicians to better understand the progression of any one patient's cancer and develop more effective treatments. The area of cancer genomics and evolution has the potential to revolutionize understanding of cancer and transform patient care.
Single-cell sequencing provides an unprecedented resolution of activity at the cellular level. Clinical and basic researchers have been increasingly using single-cell genome, methylome and transcriptome sequencing to better understand disease mechanisms. Researchers in the division are active in both applied and methodological research of single-cell genomics. They strive to establish the best practices for single-cell sequencing data analysis and develop computational pipelines to streamline the analysis. Using the recent developments in high-dimensional statistics and deep learning, they are developing quantitative tools to meet downstream analytical needs, including differential abundance and expression analysis, multi-omics integration, and predictive modeling.
Proteomics and metabolomics
The proteome and metabolome together are the main mediators between a biological system's genotype and its phenotype. They are also key carriers of a system's response to external stimuli. Hence, studying proteins and metabolites and their dynamics is key to understanding a biological system. To this end, proteomics and metabolomics offer an array of high-throughput "omics" tools to measure the expression and modification of proteins and metabolites present in a system. Computational biology faculty members are experts in analyzing proteomic and metabolomic data to find molecules and pathways that are differentially expressed or differentially modified between experimental conditions or patient groups. Our faculty has expertise in using novel algorithms to analyze proteomic and metabolomic data to address a wide variety of questions in the areas of clinical diagnostics, cancer phenomics, insulin-resistant diabetes, cardiovascular health, and other acute or chronic physiological conditions.
In much the same way that massively parallel sequencing technologies enabled the genomics revolution, whole-slide images are revolutionizing pathology. Whole-slide images (WSIs) are pictures captured at extremely high resolution resulting in large amounts of data (2 to 5GB). In digital pathology, these base files can be analyzed with multiple algorithms, depending on the question being asked (for example, somatic mutation, copy number variants, structural variants versus classification, segmentation, localization). The advent of WSIs and the corresponding analytics are making the qualitative discipline of pathology more quantitative.
In the Division of Computational Biology, we have the tools and skills for conversion of WSIs to digital imaging and communications in medicine (DICOM) format, classification of WSIs (tumor versus normal), nuclear detection and segmentation, digital ink removal, anomaly detection, and more. Faculty and staff are actively involved in both the development of new algorithms and the application of existing artificial intelligence (AI) and machine learning (ML) approaches for WSI analysis.