A5014298020- Gene expression and cancer classification
- Inflammatory Bowel Disease
- CAR-T cell therapy research
- RNA modifications and cancer
- Microscopic Colitis
- Monoclonal and Polyclonal Antibodies Research
- Cancer Genomics and Diagnostics
- Bioinformatics and Genomic Networks
- Cancer-related molecular mechanisms research
- Cancer Immunotherapy and Biomarkers
- Epigenetics and DNA Methylation
- Advanced Clustering Algorithms Research
- Hormonal Regulation and Hypertension
- Cancer-related Molecular Pathways
- Bayesian Methods and Mixture Models
- Ocular Oncology and Treatments
- Face and Expression Recognition
- Cell Image Analysis Techniques
- Cancer, Hypoxia, and Metabolism
- Machine Learning and Algorithms
- Lymphoma Diagnosis and Treatment
- AI in cancer detection
- Domain Adaptation and Few-Shot Learning
- Genomics and Chromatin Dynamics
- Radiomics and Machine Learning in Medical Imaging
Dana-Farber Cancer Institute
2018-2024
Harvard University
2018-2023
Roche (United States)
2021
Roche (Ireland)
2021
Janssen (United States)
2018
Washington University in St. Louis
2018
University of North Carolina at Chapel Hill
2014-2017
UNC Lineberger Comprehensive Cancer Center
2015
The Cancer Genome Atlas profiled 279 head and neck squamous cell carcinomas (HNSCCs) to provide a comprehensive landscape of somatic genomic alterations. Here we show that human-papillomavirus-associated tumours are dominated by helical domain mutations the oncogene PIK3CA, novel alterations involving loss TRAF3, amplification cycle gene E2F1. Smoking-related HNSCCs demonstrate near universal loss-of-function TP53 CDKN2A inactivation with frequent copy number including 3q26/28 11q13/22. A...
In high-throughput studies, hundreds to millions of hypotheses are typically tested. Statistical methods that control the false discovery rate (FDR) have emerged as popular and powerful tools for error control. While classic FDR use only p values input, more modern been shown increase power by incorporating complementary information informative covariates prioritize, weight, group hypotheses. However, there is currently no consensus on how compare one another. We investigate accuracy,...
Cluster analysis has proved to be an invaluable tool for the exploratory and unsupervised of high-dimensional datasets. Among methods clustering, hierarchical approaches have enjoyed substantial popularity in genomics other fields their ability simultaneously uncover multiple layers clustering structure. A critical challenging question cluster is whether identified clusters represent important underlying structure or are artifacts natural sampling variation. Few been proposed addressing this...
Immune checkpoint inhibitors (ICIs), by reinvigorating CD8+ T cell mediated immunity, have revolutionized cancer therapy. Yet, the systemic distribution, a potential biomarker of ICI response, remains poorly characterized. We assessed safety, imaging dose and timing, pharmacokinetics immunogenicity zirconium-89-labeled, CD8-specific, one-armed antibody positron emission tomography tracer 89ZED88082A in patients with solid tumors before ~30 days after starting therapy (NCT04029181). No...
Homeodomains (HDs) are the second largest class of DNA binding domains (DBDs) among eukaryotic sequence-specific transcription factors (TFs) and TF structural with number disease-associated mutations in Human Gene Mutation Database (HGMD). Despite numerous studies large-scale analyses HD specificity, HD-DNA recognition is still not fully understood. Here, we analyze 92 human mutants, including variants uncertain significance (VUS), for their effects on activity. Many alter affinity and/or...
High-throughput sequencing technologies, including RNA-seq, have made it possible to move beyond gene expression analysis study transcriptional events alternative splicing and fusions. Furthermore, recent studies in cancer suggested the importance of identifying transcriptionally altered loci as biomarkers for improved prognosis therapy. While many statistical methods been proposed novel with nearly all rely on contrasting known classes samples, such tumor normal. Few tools exist...
Cluster analysis has proved to be an invaluable tool for the exploratory and unsupervised of high dimensional datasets. Among methods clustering, hierarchical approaches have enjoyed substantial popularity in genomics other fields their ability simultaneously uncover multiple layers clustering structure. A critical challenging question cluster is whether identified clusters represent important underlying structure or are artifacts natural sampling variation. Few been proposed addressing this...
Abstract Background In high-throughput studies, hundreds to millions of hypotheses are typically tested. Statistical methods that control the false discovery rate (FDR) have emerged as popular and powerful tools for error control. While classic FDR use only p -values input, more modern been shown increase power by incorporating complementary information “informative covariates” prioritize, weight, group hypotheses. However, there is currently no consensus on how compare one another. We...
Benchmark studies are widely used to compare and evaluate tools developed for answering various biological questions. Despite the popularity of these comparisons, implementation is often ad hoc, with little consistency across studies. To address this problem, we SummarizedBenchmark, an R package framework organizing structuring benchmark comparisons. SummarizedBenchmark defines a general grammar benchmarking allows easier setup execution while improving reproducibility replicability such We...
Abstract T cell enhancing immune checkpoint inhibitors (ICI) are effective across several tumor types in a subset of patients. Insights into systemic localization cytotoxic CD8+ cells might support early treatment decisions. To address this, we performed PET imaging study with zirconium-89 (89Zr) labeled one-armed CD8-specific antibody 89ZED88082A to assess tracer performance, safety, and pharmacokinetics (PK) before during treatment. Here report preliminary data on uptake lesions ICI....
Although antibodies targeting specific tumor-expressed antigens are the standard of care for some cancers, identification cancer-specific targets amenable to antibody binding has remained a bottleneck in development new therapeutics. To overcome this challenge, we developed high-throughput platform that allows unbiased, simultaneous discovery and based on phenotypic profiles. Applying ovarian cancer, identified wide diversity cancer including receptor tyrosine kinases, adhesion migration...
Summary Homeodomains (HDs) are the second largest class of DNA binding domains (DBDs) among eukaryotic sequence-specific transcription factors (TFs) and play important roles in regulating development, body patterning, cellular differentiation. Here, we analyzed 92 human HD mutants, including disease-associated variants unknown significance (VUSs), for their effects on activity. Many altered affinity and/or specificity. Biochemical analysis structural modeling identified 14 novel...
Binary classification is a common statistical learning problem in which model estimated on set of covariates for some outcome indicating the membership one two classes. In literature, there exists distinction between hard and soft classification. classification, conditional class probability modeled as function covariates. contrast, methods only target optimal prediction boundary. While have been studied extensively, not much work has done to compare actual tasks this paper we propose...
Abstract Background Analyses of microRNA expression profiles have revealed that many microRNAs are expressed aberrantly and correlate with tumorigenesis, progression, prognosis various solid tumors. However, several miRNA analyses head neck squamous cell carcinoma (HNSCC) been inconsistent among studies due to the small sample size. Method 366 HNSCC samples from two independent subsets patients UNC The Cancer Genome Atlas (TCGA) were analyzed by different platforms, microarray Seq,...
Binary classification is a common statistical learning problem in which model estimated on set of covariates for some outcome, indicating the membership one two classes. In literature, there exists distinction between hard and soft classification. classification, conditional class probability modeled as function covariates. contrast, methods only target optimal prediction boundary. While have been studied extensively, not much work has performed to compare actual tasks this paper, we propose...