Document text
Principal Investigator: Alison Motsinger-Reif
Organization: NATIONAL INSTITUTE OF ENVIRONMENTAL HEALTH SCIENCES
Fiscal Year: 2024
Award: $912,836
Funding agency: National Institute of Environmental Health Sciences
MIBCOVIS Framework:
Led by Dr. Benedict Anchang's group, we introduced MIBCOVIS, a framework that improves the visualization and interpretation of complex biological data without needing ground truth. MIBCOVIS combines various metrics within a Bayesian model to assess data reduction methods, applied to single-cell datasets like CyTOF and scRNA-seq. Our findings highlight the importance of evaluating visualization and interpretability together, making MIBCOVIS a valuable tool for analyzing dynamic and spatial biological processes.
ToxPi Enhancement:
Led by Dr. David Reif's group, enhanced the Toxicological Prioritization Index (ToxPi), a tool for visualizing and ranking chemical toxicity, expanding its application to geospatial analyses like COVID-19 risk and PFAS exposure. To improve the accuracy of ToxPi's scoring, we developed a method using ordinal regression and a genetic algorithm to better predict feature weights. This approach improves weight prediction and sample ranking, increasing ToxPi’s accuracy and usability.
Machine Learning and Complex Survey Data:
Along with Dr. Nat McNell, we explored the challenges of using machine learning in epidemiologic research, particularly with complex survey data. Analyzing data from NHANES, we found that ignoring sampling weights in gradient boosting models led to overestimated performance. Recalculating with weighted outcomes improved accuracy. Our findings underscore the importance of using sampling weights to ensure accurate predictions in complex surveys.
ToxicR Development:
ALong with Dr. Matt Wheeler, we developed ToxicR, an open-source R package for computational toxicology, to address the need for more flexible tools. ToxicR, created by NIEHS in collaboration with NTP and EPA, includes standard analyses like dose-response and trend tests while allowing users to program new algorithms. Built on the same codebase as EPA and NTP software, ToxicR offers customizable workflows, making it adaptable for future toxicogenomic research.
Ongoing projects building onto methods for detecting gene-environment interactions are currently ongoing, using variance QTLs to prioritize single nucleotide polymorphisms for detecting gene-gene interactions.
Additionally, Dr. Ziyue Wang is working on developing new normalization approaches for microbiome data.
Terms: <Address><Algorithms><Bayesian Modeling><Bayesian adaptive designs><Bayesian adaptive models><Bayesian belief network><Bayesian belief updating model><Bayesian framework><Bayesian hierarchical model><Bayesian network model><Bayesian nonparametric models><Bayesian spatial data model><Bayesian spatial image models><Bayesian spatial models><Bayesian statistical models><Bayesian tracking algorithms><Biological><Biological Function><Biological Process><COVID-19 related risk><COVID-19 risk><COVID-19 risk factor><Cell Body><Cells><Chemicals><Code><Coding System><Collaborations><Complex><Computer software><Data><Data Set><Development><Dose><Ensure><Epidemiologic Research><Epidemiologic Studies><Epidemiological Studies><Epidemiology Research><Future><GWA study><GWAS><Gene x Environment Interaction><Genes><Genetic Algorithm><Genetic Programming><Genomics><GxE interaction><Machine Learning><Methods><Modeling><NHANES><NIEHS><National Health and Nutrition Examination Survey><National Institute of Environmental Health Sciences><PFAS><Performance><Poly-fluoroalkyl substances><Polymorphism Analysis><Polymorphism Detection><QTL><Quantitative Trait Loci><Research><Sampling><Single Base Polymorphism><Single Nucleotide Polymorphism><Software><Survey Instrument><Surveys><Testing><Toxic effect><Toxicities><Toxicogenomics><Toxicology><Visualization><Visualization software><Weight><Work><base><bases><biologic><computational toxicology><coronavirus disease 2019 risk><coronavirus disease 2019 risk factor><data reduction><detection method><detection procedure><detection technique><developmental><environment effect on gene><epidemiologic investigation><epidemiology study><flexibility><flexible><gene environment interaction><gene interaction><genome wide association><genome wide association scan><genome wide association studies><genome wide association study><genomewide association scan><genomewide association studies><genomewide association study><gradient boosting><high dimensionality><improved><improved outcome><indexing><machine based learning><metabolism measurement><metabolomics><metabonomics><method development><microbiome><open source><perfluorinated alkyl substances><perfluoroalkyl substances><perfluoroalkylated substances><polyfluorinated alkyl substances><polyfluoroalkyl substances><programs><response><risk associated with COVID-19><risk factor associated with COVID-19><risk factor related to COVID-19><risk related to COVID-19><scRNA-seq><single cell RNA-seq><single cell RNAseq><single cell expression profiling><single cell transcriptomic profiling><single nucleotide variant><single-cell RNA sequencing><tool><trend><usability><visualization tool><weights><whole genome association analysis><whole genome association studies><whole genome association study>