Publications

Frontier artificial intelligence (AI) models have advanced rapidly through training on internet-scale public data, yet such systems lack access to private clinical data. Neuroimaging is underrepresented in the public domain due to identifiable facial features within magnetic resonance imaging (MRI) and computed tomography (CT) scans, restricting model performance in clinical medicine. Here we show that frontier models underperform on neuroimaging tasks and that learning directly from uncurated data generated during routine clinical care at health systems, a paradigm we call 'health system learning', yields high-performance, generalist neuroimaging models. We introduce NeuroVFM, a visual foundation model trained on 5.24 million clinical MRI and CT volumes using a scalable volumetric predictive architecture. NeuroVFM learns comprehensive representations of brain anatomy and pathology, achieving state-of-the-art performance across multiple clinical tasks, including radiologic diagnosis and report generation. The model embeds MRI and CT scans into a shared neuroanatomic latent space and grounds diagnostic findings. When paired with open-source language models, NeuroVFM generates radiology reports that surpass frontier models in accuracy, clinical triage and expert preference. NeuroVFM reduces hallucinated findings and critical errors, offering safer clinical decision support. These results establish health system learning as a paradigm for building generalist medical AI and provide a scalable framework for clinical foundation models.

Neuroimaging is a ubiquitous tool for evaluating patients with neurological diseases. The global demand for magnetic resonance imaging (MRI) studies has risen steadily, placing substantial strain on health systems, prolonging turnaround times and intensifying physician burnout. These challenges disproportionately impact patients in low-resource and rural settings. Here we utilize data from a large academic health system to develop Prima, an AI foundation model for neuroimaging that supports real-world, clinical MRI studies as input. Trained on over 220,000 MRI studies, Prima uses a hierarchical vision architecture that provides general and transferable MRI features. Prima was tested in a 1-year health system-wide study that included 29,431 MRI studies. Across 52 radiologic diagnoses from major neurologic disorders, Prima achieved a mean diagnostic area under the curve (AUC) of 92.0%, outperforming other state-of-the-art general and medical AI models. Prima offers explainable differential diagnoses, worklist priority for radiologists and clinical referral recommendations. Prima demonstrates algorithmic fairness across sensitive groups. These findings highlight the transformative potential of health system-scale AI training and Prima's role in advancing AI-driven healthcare.

Vaccines are widely used in both research and clinical settings. To facilitate FAIR data practices, we urgently need to standardize vaccine representation, integrate information across diverse vaccine types, and support computer-assisted reasoning. Accordingly, we have since 2007 developed the community-based Vaccine Ontology (VO), which aligns with the Basic Formal Ontology and adheres to OBO Foundry principles. VO ontologically models vaccines, vaccine components, vaccine immune responses, vaccine investigation studies and other vaccine-related topics. VO represents more than 10,000 vaccines targeting 289 infectious pathogens and cancers in humans and over 30 nonhuman animal species. VO provides mappings to external resources such as RxNorm, CVX, FDA, and USDA. VO facilitates vaccine standardization in resources such as the VIOLIN vaccine database, ImmPort, and the Vaccine Adjuvant Compendium (VAC). VO enables semantic queries on vaccine data. It has been shown to enhance the analysis of experimental and clinical vaccine datasets, as well as vaccine-related literature mining. Overall, VO standardizes vaccine modeling and representation and greatly supports vaccine AI research in the Semantic Web era.

In multivariate longitudinal studies, associations between outcomes often exhibit time-varying and individual level heterogeneity, motivating the modeling of correlations as an explicit function of time and covariates. However, most existing methods for correlation analysis fail to simultaneously capture the time-varying and covariate-dependent effects. We propose a Time-Varying and Covariate-Dependent (TiVAC) correlation model that jointly allows covariate effects on correlation to change flexibly and smoothly across time. TiVAC employs a bivariate Gaussian model where the covariate-dependent correlations are modeled semiparametrically using penalized splines. We develop a penalized maximum likelihood-based Newton-Raphson algorithm, and inference on time-varying effects is provided through simultaneous confidence bands. Simulation studies show that TiVAC consistently outperforms existing methods in accurately estimating correlations across a wide range of settings, including binary and continuous covariates, sparse to dense observation schedules, and across diverse correlation trajectory patterns. We apply TiVAC to a psychiatric case study of 291 bipolar I patients, modeling the time-varying correlation between depression and anxiety scores as a function of their clinical variables. Our analyses reveal significant heterogeneity associated with gender and nervous-system medication use, which varies with age, revealing the complex dynamic relationship between depression and anxiety in bipolar disorders.

With the increasing volume of biomedical experimental data, standardizing, sharing, and integrating heterogeneous experimental data across domains has become a major challenge. To address this challenge, we have developed an ontology-supported Study-Experiment-Assay (SEA) common data model (CDM), which includes 10 core and 3 auxiliary classes based on object-oriented modeling. SEA CDM uses interoperable ontologies for data standardization and knowledge inference. Building on the SEA CDM, we developed the Ontology-based SEA Network (OSEAN) relational database and knowledge graph, along with a set of ETL (Extract, Transform, Load) and query tools, and further applied them to represent 1,278 immune studies with over two million samples from three resources: VIGET, ImmPort, and CELLxGENE. Using simple, robust queries and analyses, our research identified multiple scientific insights into sex-specific immune responses, such as neutrophil degranulation and TNF binding to physiological receptors, following live attenuated and trivalent inactivated influenza vaccination. The novel SEA CDM system lays a foundation for establishing an integrative biodata ecosystem across biological and biomedical domains.

Sample size calculations for power analysis are critical for clinical research and trial design, yet their complexity and reliance on statistical expertise create barriers for many researchers. We introduce PowerGPT, an AI-powered system integrating large language models (LLMs) with statistical engines to automate test selection and sample size estimation in trial design. In a randomized trial to evaluate its effectiveness, PowerGPT significantly improved task completion rates (99.3% vs. 88.9% for test selection, 99.3% vs. 77.8% for sample size calculation) and accuracy (94.1% vs. 55.4% in sample size estimation, p < 0.001), while reducing average completion time (4.0 vs. 9.3 minutes, p < 0.001). These gains were consistent across various statistical tests and benefited both statisticians and non-statisticians as well as bridging expertise gaps. Already under deployment across multiple institutions, PowerGPT represents a scalable AI-driven approach that enhances accessibility, efficiency, and accuracy in statistical power analysis for clinical research.

BACKGROUND: Bipolar disorder is a complex psychiatric condition characterized by recurrent mood episodes and fluctuating symptom severity. Our study investigates post-diagnosis complexities using 17-year longitudinal data from the Prechter Bipolar Research Program. METHODS: We analyzed mood trajectories of 525 bipolar I disorder (BP1) and 150 bipolar II disorder (BP2) patients, using the Linear Analysis Mixed Model Phenotyper (LAMP) to identify subgroups. This approach revealed shared and distinct patterns across individual trajectories, classifying these subgroups into meaningful longitudinal phenotypes. Regression modeling was then applied to identify associations between these phenotypes and relevant factors. RESULTS: We found that medication use and alcohol use disorder significantly impacted long-term mood trajectories. Polypharmacy was associated with resistant/worsening trajectories in both BP1 (PHQ-9: OR: 1.09, CI: 1.02-1.17) and BP2 (GAD-7: OR: 1.21, CI: 1.05-1.43), with each additional medication increasing the risk of a worse trajectory by 9% and 21%, respectively. In BP1, antidepressant use improved trajectories (PHQ-9: OR: 0.40, CI: 0.19-0.83), but when combined with antipsychotics, increasing use frequency from 0% to 20% raised the risk of resistant/worsening trajectories (OR: 2.09, CI: 1.33- 3.28). In BP2, antipsychotic use was associated with improved trajectories (PHQ-9: OR: 0.04, CI: 0.00-0.56). Alcohol use disorder in BP1 increased resistant/worsening risk by 83% (PHQ-9: OR: 1.83, CI: 1.15-2.93). CONCLUSIONS: Our results indicate that polypharmacy and alcohol use disorder contribute to resistant or worsening mood trajectories. In BP1, while antidepressants are beneficial, their combination with antipsychotics substantially increases risk, highlighting the need for individualized, diagnosis-specific treatment approaches. HIGHLIGHTS: LAMP analysis pipeline aided in improving treatment approaches in bipolar disorder.17 years of PrBP data provided comprehensive analysis of long-term mood trajectories.Polypharmacy and alcohol use disorder worsened mood trajectory.The combination of antidepressants and antipsychotics worsened mood trajectory.

Microscopy enables direct observation of cellular morphology in 3D, with transmitted-light methods offering low-cost, minimally invasive imaging and fluorescence microscopy providing specificity and contrast. Virtual staining combines these strengths by using machine learning to predict fluorescence images from label-free inputs. However, training of existing methods typically relies on loss functions that treat all pixels equally, thus reproducing background noise and artifacts instead of focusing on biologically meaningful signals. We introduce Spotlight, a simple yet powerful virtual staining approach that guides the model to focus on relevant cellular structures. Spotlight uses histogram-based foreground estimation to mask pixel-wise loss and to calculate a Dice loss on soft-thresholded predictions for shape-aware learning. Applied to a 3D benchmark dataset, Spotlight improves morphological representation while preserving pixel-level accuracy, resulting in virtual stains better suited for downstream tasks such as segmentation and profiling.

Over the past few years, artificial intelligence (AI) has emerged as a transformative force in drug discovery and development (DDD), revolutionizing many aspects of the process. This survey provides a comprehensive review of recent advancements in AI applications within early drug discovery and post-market drug assessment. It addresses the identification and prioritization of new therapeutic targets, prediction of drug-target interaction (DTI), design of novel drug-like molecules, and assessment of the clinical efficacy of new medications. By integrating AI technologies, pharmaceutical companies can accelerate the discovery of new treatments, enhance the precision of drug development, and bring more effective therapies to market. This shift represents a significant move towards more efficient and cost-effective methodologies in the DDD landscape.

Despite implementing hundreds of strategies, cancer drug development suffers from a 95% failure rate over 30 years, with only 30% of approved cancer drugs extending patient survival beyond 2.5 months. Adding more criteria without eliminating nonessential ones is impractical and may fall into the "survivorship bias" trap. Machine learning (ML) models may enhance efficiency by saving time and cost. Yet, they may not improve success rate without identifying the root causes of failure. We propose a "STAR-guided ML system" (structure-tissue/cell selectivity-activity relationship) to enhance success rate and efficiency by addressing three overlooked interdependent factors: potency/specificity to the on/off-targets determining efficacy in tumors at clinical doses, on/off-target-driven tissue/cell selectivity influencing adverse effects in the normal organs at clinical doses, and optimal clinical doses balancing efficacy/safety as determined by potency/specificity and tissue/cell selectivity. STAR-guided ML models can directly predict clinical dose/efficacy/safety from five features to design/select the best drugs, enhancing success and efficiency of cancer drug development.

Understanding the functional effects of sequence variation is crucial in genomics. Individual human genomes contain millions of variants that contribute to phenotypic variability and disease risks at the population level. Because variants rarely act in isolation, we must consider potential interactions of neighboring variants to accurately predict functional effects. We can accomplish this using haplotagging, which matches sequencing reads to their parental haplotypes using alleles observed at known heterozygous variants. However, few published tools for haplotagging exist and these share several technical and usability-related shortcomings that limit applicability, in particular a lack of insight or control over error rates, and lack of key metrics on the underlying sources of haplotagging error. Here we present HaplotagLR: a user-friendly tool that haplotags long sequencing reads based on a multinomial model and existing phased variant lists. HaplotagLR is user-configurable and includes a basic error model to control the empirical FDR in its output. We show that HaplotagLR outperforms the leading haplotagging method in simulated datasets, especially at high levels of specificity, and displays 7% greater sensitivity in haplotagging real data. HaplotagLR advances both the immediate utility of haplotagging and paves the way for further improvements to this important method.

COVID-19 often manifests with different outcomes in different patients, highlighting the complexity of the host-pathogen interactions involved in manifestations of the disease at the molecular and cellular levels. In this paper, we propose a set of postulates and a framework for systematically understanding complex molecular host-pathogen interaction networks. Specifically, we first propose four host-pathogen interaction (HPI) postulates as the basis for understanding molecular and cellular host-pathogen interactions and their relations to disease outcomes. These four postulates cover the evolutionary dispositions involved in HPIs, the dynamic nature of HPI outcomes, roles that HPI components may occupy leading to such outcomes, and HPI checkpoints that are critical for specific disease outcomes. Based on these postulates, an HPI Postulate and Ontology (HPIPO) framework is proposed to apply interoperable ontologies to systematically model and represent various granular details and knowledge within the scope of the HPI postulates, in a way that will support AI-ready data standardization, sharing, integration, and analysis. As a demonstration, the HPI postulates and the HPIPO framework were applied to study COVID-19 with the Coronavirus Infectious Disease Ontology (CIDO), leading to a novel approach to rational design of drug/vaccine cocktails aimed at interrupting processes occurring at critical host-coronavirus interaction checkpoints. Furthermore, the host-coronavirus protein-protein interactions (PPIs) relevant to COVID-19 were predicted and evaluated based on prior knowledge of curated PPIs and domain-domain interactions, and how such studies can be further explored with the HPI postulates and the HPIPO framework is discussed.

BACKGROUND: The current COVID-19 pandemic and the previous SARS/MERS outbreaks of 2003 and 2012 have resulted in a series of major global public health crises. We argue that in the interest of developing effective and safe vaccines and drugs and to better understand coronaviruses and associated disease mechenisms it is necessary to integrate the large and exponentially growing body of heterogeneous coronavirus data. Ontologies play an important role in standard-based knowledge and data representation, integration, sharing, and analysis. Accordingly, we initiated the development of the community-based Coronavirus Infectious Disease Ontology (CIDO) in early 2020. RESULTS: As an Open Biomedical Ontology (OBO) library ontology, CIDO is open source and interoperable with other existing OBO ontologies. CIDO is aligned with the Basic Formal Ontology and Viral Infectious Disease Ontology. CIDO has imported terms from over 30 OBO ontologies. For example, CIDO imports all SARS-CoV-2 protein terms from the Protein Ontology, COVID-19-related phenotype terms from the Human Phenotype Ontology, and over 100 COVID-19 terms for vaccines (both authorized and in clinical trial) from the Vaccine Ontology. CIDO systematically represents variants of SARS-CoV-2 viruses and over 300 amino acid substitutions therein, along with over 300 diagnostic kits and methods. CIDO also describes hundreds of host-coronavirus protein-protein interactions (PPIs) and the drugs that target proteins in these PPIs. CIDO has been used to model COVID-19 related phenomena in areas such as epidemiology. The scope of CIDO was evaluated by visual analysis supported by a summarization network method. CIDO has been used in various applications such as term standardization, inference, natural language processing (NLP) and clinical data integration. We have applied the amino acid variant knowledge present in CIDO to analyze differences between SARS-CoV-2 Delta and Omicron variants. CIDO's integrative host-coronavirus PPIs and drug-target knowledge has also been used to support drug repurposing for COVID-19 treatment. CONCLUSION: CIDO represents entities and relations in the domain of coronavirus diseases with a special focus on COVID-19. It supports shared knowledge representation, data and metadata standardization and integration, and has been used in a range of applications.

Histone deacetylase inhibitors, such as valproic acid (VPA), have important clinical therapeutic and cellular reprogramming applications. They induce chromatin reorganization that is associated with altered cellular morphology. However, there is a lack of comprehensive characterization of VPA-induced changes of nuclear size and shape. Here, we quantify 3D nuclear morphology of primary human astrocyte cells treated with VPA over time (hence, 4D). We compared volumetric and surface-based representations and identified seven features that jointly discriminate between normal and treated cells with 85% accuracy on day 7. From day 3, treated nuclei were more elongated and flattened and then continued to morphologically diverge from controls over time, becoming larger and more irregular. On day 7, most of the size and shape descriptors demonstrated significant differences between treated and untreated cells, including a 24% increase in volume and 6% reduction in extent (shape regularity) for treated nuclei. Overall, we show that 4D morphometry can capture how chromatin reorganization modulates the size and shape of the nucleus over time. These nuclear structural alterations may serve as a biomarker for histone (de-)acetylation events and provide insights into mechanisms of astrocytes-to-neurons reprogramming.

Acute respiratory distress syndrome (ARDS) is a life-threatening lung injury with global prevalence and high mortality. Chest x-rays (CXR) are critical in the early diagnosis and treatment of ARDS. However, imaging findings may not result in proper identification of ARDS due to a number of reasons, including nonspecific appearance of radiological features, ambiguity in a patient's case due to the pathological stage of the disease, and poor inter-rater reliability from interpretations of CXRs by multiple clinical experts. This study demonstrates the potential capability of methodologies in artificial intelligence, machine learning, and image processing to overcome these challenges and quantitatively assess CXRs for presence of ARDS. We propose and describe Directionality Measure, a novel feature engineering technique used to capture the "cloud-like" appearance of diffuse alveolar damage as a mathematical concept. This study also examines the effectiveness of using an off-the-shelf, pretrained deep learning model as a feature extractor in addition to standard features extracted from the histogram and gray-level co-occurrence matrix (GLCM). Data was collected from hospitalized patients at Michigan Medicine's intensive care unit and the cohort's inclusion criteria was specifically designed to be representative of patients at risk of developing ARDS. Multiple machine learning models were used to evaluate these features with 5-fold cross-validation and the final performance was reported on a hold-out, temporally distinct test set. With AdaBoost, Directionality Measure achieved an accuracy of 78% and AUC of 74% - outperforming classification results using features from the histogram (75% accuracy and 73% AUC), GLCM (76% accuracy and 73% AUC), and ResNet-50 (77% accuracy and 73% AUC). Further experimental results demonstrated that using all feature sets in combination achieved the best overall performance, yielding an accuracy of 83% and AUC of 79% with AdaBoost. These results demonstrate the potential capability of using the proposed methodologies to complement current clinical analysis for detection of ARDS from CXRs.

BACKGROUND: This study outlines an image processing algorithm for accurate and consistent lung segmentation in chest radiographs of critically ill adults and children typically obscured by medical equipment. In particular, this work focuses on applications in analysis of acute respiratory distress syndrome - a critical illness with a mortality rate of 40% that affects 200,000 patients in the United States and 3 million globally each year. METHODS: Chest radiographs were obtained from critically ill adults (n = 100), adults diagnosed with acute respiratory distress syndrome (ARDS) (n = 25), and children (n = 100) hospitalized at Michigan Medicine. Physicians annotated the lung field of each radiograph to establish the ground truth. A Total Variation-based Active Contour (TVAC) lung segmentation algorithm was developed and compared to multiple state-of-the-art methods including a deep learning model (U-Net), a random walker algorithm, and an active spline model, using the Sørensen-Dice coefficient to measure segmentation accuracy. RESULTS: The TVAC algorithm accurately segmented lung fields in all patients in the study. For the adult cohort, an averaged Dice coefficient of 0.86 ±0.04 (min: 0.76) was reported for TVAC, 0.89 ±0.12 (min: 0.01) for U-Net, 0.74 ±0.19 (min: 0.15) for the random walker algorithm, and 0.64 ±0.17 (min: 0.20) for the active spline model. For the pediatric cohort, a Dice coefficient of 0.85 ±0.04 (min: 0.75) was reported for TVAC, 0.87 ±0.09 (min: 0.56) for U-Net, 0.67 ±0.18 (min: 0.18) for the random walker algorithm, and 0.61 ±0.18 (min: 0.18) for the active spline model. CONCLUSION: The proposed algorithm demonstrates the most consistent performance of all segmentation methods tested. These results suggest that TVAC can accurately identify lung fields in chest radiographs in critically ill adults and children.

The pharmacogenomic network responsible for the rapid antidepressant action of ketamine and concomitant adverse events in patients has been poorly defined. Integrative, multi-scale biological data analytics helps explain ketamine’s action. Using a validated computational pipeline, candidate ketamine-response genes and regulatory RNAs from published literature, binding affinity studies, and single nucleotide polymorphisms (SNPs) from genomewide association studies (GWAS), we identified 108 SNPs associated with 110 genes and regulatory RNAs. All of these SNPs are classified as enhancers, and additional chromatin interaction mapping in human neural cell lines and tissue shows enhancer-promoter interactions involving other network members. Pathway analysis and gene set optimization identified three composite sub-networks within the broader ketamine pharmacogenomic network. Expression patterns of ketamine network genes within the postmortem human brain are concordant with ketamine neurocircuitry based on the results of 24 published functional neuroimaging studies. The ketamine pharmacogenomic network is enriched in forebrain regions known to be rapidly activated by ketamine, including cingulate cortex and frontal cortex, and is significantly regulated by ketamine ( p =6.26E-33; Fisher’s exact test). The ketamine pharmacogenomic network can be partitioned into distinct enhancer sub-networks associated with: (1) glutamate neurotransmission, chromatin remodeling, smoking behavior, schizophrenia, pain, nausea, vomiting, and post-operative delirium; (2) neuroplasticity, depression, and alcohol consumption; and (3) pharmacokinetics. The component sub-networks explain the diverse action mechanisms of ketamine and its analogs. These results may be useful for optimizing pharmacotherapy in patients diagnosed with depression, pain or related stress disorders. One Sentence Summary The ketamine network in the human brain consists of sub-networks associated with glutamate neurotransmission, neuroplasticity, and pharmacokinetics.

Chromosome conformation capture methods have revealed the dynamics of genome architecture which is spatially organized into topologically associated domains, with gene regulation mediated by enhancer-promoter pairs in chromatin space. New evidence shows that endogenous hormones and several xenobiotics act within circumscribed topological domains of the spatial genome, impacting subsets of the chromatin contacts of enhancer-gene promoter pairs in cis and trans Results from the National Institutes of Health-funded PsychENCODE project and the study of chromatin remodeling complexes have converged to provide a clearer understanding of the organization of the neurogenic epigenome in humans. Neuropsychiatric diseases, including schizophrenia, bipolar spectrum disorder, autism spectrum disorder, attention deficit hyperactivity disorder, and other neuropsychiatric disorders are significantly associated with mutations in neurogenic transcriptional networks. In this review, we have reanalyzed the results from publications of the PsychENCODE Consortium using pharmacoinformatics network analysis to better understand druggable targets that control neurogenic transcriptional networks. We found that valproic acid and other psychotropic drugs directly alter these networks, including chromatin remodeling complexes, transcription factors, and other epigenetic modifiers. We envision a new generation of CNS therapeutics targeted at neurogenic transcriptional control networks, including druggable parts of chromatin remodeling complexes and master transcription factor-controlled pharmacogenomic networks. This may provide a route to the modification of interconnected gene pathways impacted by disease in patients with neuropsychiatric and neurodegenerative disorders. Direct and indirect therapeutic strategies to modify the master regulators of neurogenic transcriptional control networks may ultimately help extend the life span of CNS neurons impacted by disease.

Quantitative analysis of morphological changes in a cell nucleus is important for the understanding of nuclear architecture and its relationship with pathological conditions such as cancer. However, dimensionality of imaging data, together with a great variability of nuclear shapes, presents challenges for 3D morphological analysis. Thus, there is a compelling need for robust 3D nuclear morphometric techniques to carry out population-wide analysis. We propose a new approach that combines modeling, analysis, and interpretation of morphometric characteristics of cell nuclei and nucleoli in 3D. We used robust surface reconstruction that allows accurate approximation of 3D object boundary. Then, we computed geometric morphological measures characterizing the form of cell nuclei and nucleoli. Using these features, we compared over 450 nuclei with about 1,000 nucleoli of epithelial and mesenchymal prostate cancer cells, as well as 1,000 nuclei with over 2,000 nucleoli from serum-starved and proliferating fibroblast cells. Classification of sets of 9 and 15 cells achieved accuracy of 95.4% and 98%, respectively, for prostate cancer cells, and 95% and 98% for fibroblast cells. To our knowledge, this is the first attempt to combine these methods for 3D nuclear shape modeling and morphometry into a highly parallel pipeline workflow for morphometric analysis of thousands of nuclei and nucleoli in 3D.

The search for causative pharmacogenomic loci is being transformed by integrative omics pipelines, but their outputs have only begun being applied to test design. We assess the direction of the field in light of Biobanks/PheWAS, omics atlases, and AI. We first assess the potential of recent epigenome and spatial genome concepts, datasets, and methods to improve the functionality of PIP-style pipelines. We then discuss new potential methods of genetic test design on the basis of the outputs of such pipelines. We conclude with a vision for a pharmacophenomic atlas, in which omics atlas data, PheWAS associations, and biobank data would be used with AI to design thousands of genetic tests for clinical deployment in an automated parallel process.

Changes in cell nuclear architecture are regulated by complex biological mechanisms that associated with the altered functional properties of a cell. Quantitative analyses of structural alterations of nuclei and their compartments are important for understanding such mechanisms. In this work we present a comparison of approaches for nuclear structure classification, evaluated on 2D per-channel representations from a 3D microscopy imaging dataset by maximum intensity projection. Specifically, we compare direct classification of pixel data from either raw intensity images or binary masks that contain only information about morphology of the object, but not intensity. We evaluate a number of widely used classification algorithms using 2 different cross-validation schemes to assess batch effects. We compare obtained results with the previously reported baselines and discuss novel findings.

This Perspective provides examples of current and future applications of deep learning in pharmacogenomics, including: identification of novel regulatory variants located in noncoding domains of the genome and their function as applied to pharmacoepigenomics; patient stratification from medical records; and the mechanistic prediction of drug response, targets and their interactions. Deep learning encapsulates a family of machine learning algorithms that has transformed many important subfields of artificial intelligence over the last decade, and has demonstrated breakthrough performance improvements on a wide range of tasks in biomedicine. We anticipate that in the future, deep learning will be widely used to predict personalized drug response and optimize medication selection and dosing, using knowledge extracted from large and complex molecular, epidemiological, clinical and demographic datasets.

AIM: 'Pharmacoepigenomics' methods informed by omics datasets and pre-existing knowledge have yielded discoveries in neuropsychiatric pharmacogenomics. Now we evaluate the generality of these methods by discovering an extended warfarin pharmacogenomics pathway. MATERIALS & METHODS: We developed the pharmacoepigenomics informatics pipeline, a scalable multi-omics variant screening pipeline for pharmacogenomics, and conducted an experiment in the genomics of warfarin. RESULTS: We discovered known and novel pharmacogenomics variants and genes, both coding and regulatory, for warfarin response, including adverse events. Such genes and variants cluster in a warfarin response pathway consolidating known and novel warfarin response variants and genes. CONCLUSION: These results can inform a new warfarin test. The pharmacoepigenomics informatics pipeline may be able to discover new pharmacogenomics markers in other drug-disease systems.

The pharmacoepigenome can be defined as the active, noncoding province of the genome including canonical spatial and temporal regulatory mechanisms of gene regulation that respond to xenobiotic stimuli. Many psychotropic drugs that have been in clinical use for decades have ill-defined mechanisms of action that are beginning to be resolved as we understand the transcriptional hierarchy and dynamics of the nucleus. In this review, we describe spatial, temporal and biomechanical mechanisms mediated by psychotropic medications. Focus is placed on a bioinformatics pipeline that can be used both for detection of pharmacoepigenomic variants that discretize drug response and adverse events to improve pharmacogenomic testing, and for the discovery of novel CNS therapeutics. This approach integrates the functional topology and dynamics of the transcriptional hierarchy of the pharmacoepigenome, gene variant-driven identification of pharmacogenomic regulatory domains, and mesoscale mapping for the discovery of novel CNS pharmacodynamic pathways in human brain. Examples of the application of this pipeline are provided, including the discovery of valproic acid (VPA) mediated transcriptional reprogramming of neuronal cell fate following injury, and mapping of a CNS pathway glutamatergic pathway for the mood stabilizer lithium. These examples in regulatory pharmacoepigenomics illustrate how ongoing research using the 4D nucleome provides a foundation to further insight into previously unrecognized psychotropic drug pharmacodynamic pathways in the human CNS.

OBJECTIVES: To determine the mechanism of action of valproic acid (VPA) in the adult central nervous system (CNS) following traumatic brain injury (TBI) and hemorrhagic shock (HS). METHODS: Data were analyzed from different sources, including experiments in a porcine model, data from postmortem human brain, published studies, public and commercial databases. RESULTS: The transcriptional program in the CNS following TBI, HS, and VPA treatment includes activation of regulatory pathways that enhance neurogenesis and suppress gliogenesis. Genes which encode the transcription factors (TFs) that specify neuronal cell fate, including MEF2D, MYT1L, NEUROD1, PAX6 and TBR1, and their target genes, are induced by VPA. VPA represses genes responsible for oligodendrogenesis, maintenance of white matter, T-cell activation, angiogenesis, and endothelial cell proliferation, adhesion and chemotaxis. NEUROD1 has regulatory interactions with 38% of the genes regulated by VPA in a swine model of TBI and HS in adult brain. Hi-C spatial mapping of a VPA pharmacogenomic SNP in the GRIN2B gene shows it is part of a transcriptional hub that contacts 12 genes that mediate chromatin-mediated neurogenesis and neuroplasticity. CONCLUSIONS: Following TBI and HS, this study shows that VPA administration acts in the adult brain through differential activation of TFs responsible for neurogenesis, genes responsible for neuroplasticity, and repression of TFs that specify oligodendrocyte cell fate, endothelial cell chemotaxis and angiogenesis. Short title: Mechanism of action of valproic acid in traumatic brain injury.

INTRODUCTION: A translational bioinformatics challenge exists in connecting population and individual clinical phenotypes in various formats to biological mechanisms. The Medical Dictionary for Regulatory Activities (MedDRA(®)) is the default dictionary for adverse event (AE) reporting in the US Food and Drug Administration Adverse Event Reporting System (FAERS). The ontology of adverse events (OAE) represents AEs as pathological processes occurring after drug exposures. OBJECTIVES: The aim of this work was to establish a semantic framework to link biological mechanisms to phenotypes of AEs by combining OAE with MedDRA(®) in FAERS data analysis. We investigated the AEs associated with tyrosine kinase inhibitors (TKIs) and monoclonal antibodies (mAbs) targeting tyrosine kinases. The five selected TKIs/mAbs (i.e., dasatinib, imatinib, lapatinib, cetuximab, and trastuzumab) are known to induce impaired ventricular function (non-QT) cardiotoxicity. RESULTS: Statistical analysis of FAERS data identified 1053 distinct MedDRA(®) terms significantly associated with TKIs/mAbs, where 884 did not have corresponding OAE terms. We manually annotated these terms, added them to OAE by the standard OAE development strategy, and mapped them to MedDRA(®). The data integration to provide insights into molecular mechanisms of drug-associated AEs was performed by including linkages in OAE for all related AE terms to MedDRA(®) and the existing ontologies, including the human phenotype ontology (HP), Uber anatomy ontology (UBERON), and gene ontology (GO). Sixteen AEs were shared by all five TKIs/mAbs, and each of 17 cardiotoxicity AEs was associated with at least one TKI/mAb. As an example, we analyzed "cardiac failure" using the relations established in OAE with other ontologies and demonstrated that one of the biological processes associated with cardiac failure maps to the genes associated with heart contraction. CONCLUSION: By expanding the existing OAE ontological design, our TKI use case demonstrated that the combination of OAE and MedDRA(®) provides a semantic framework to link clinical phenotypes of adverse drug events to biological mechanisms.

The 4D nucleome has the potential to render challenges in neuropsychiatric pharmacogenomics more tractable. The epigenome roadmap consortium has demonstrated the critical role that noncoding regions of the human genome play in determination of human phenotype. Chromosome conformation capture methods have revealed the 4D organization of the nucleus, bringing interactions between distant regulatory elements into close spatial proximity in a periodic manner. These functional interactions have the potential to elucidate mechanisms of CNS drug response and side effects that previously have been unrecognized. This perspective assesses recent advances likely to reveal novel pharmacodynamic regulatory pathways in human brain, charting a future new avenue of pharmacogenomics research, using the spatial and temporal architecture of the human epigenome as its foundation.

AIM: To provide insight into potential regulatory mechanisms of gene expression underlying addiction, analgesia, psychotropic drug response and adverse drug events, genome-wide association studies searching for variants associated with these phenotypes has been undertaken with limited success. We undertook analysis of these results with the aim of applying epigenetic knowledge to aid variant discovery and interpretation. METHODS: We applied conditional imputation to results from 26 genome-wide association studies and three candidate gene-association studies. The analysis workflow included data from chromatin conformation capture, chromatin state annotation, DNase I hypersensitivity, hypomethylation, anatomical localization and biochronicity. We also made use of chromatin state data from the epigenome roadmap, transcription factor-binding data, spatial maps from published Hi-C datasets and 'guilt by association' methods. RESULTS: We identified 31 pharmacoepigenomic SNPs from a total of 2024 variants in linkage disequilibrium with lead SNPs, of which only 6% were coding variants. Interrogation of chromatin state using our workflow and the epigenome roadmap showed agreement on 34 of 35 tissue assignments to regulatory elements including enhancers and promoters. Loop boundary domains were inferred by association with CTCF (CCCTC-binding factor) and cohesin, suggesting proximity to topologically associating domain boundaries and enhancer clusters. Spatial interactions between enhancer-promoter pairs detected both known and previously unknown mechanisms. Addiction and analgesia SNPs were common in relevant populations and exhibited large effect sizes, whereas a SNP located in the promoter of the SLC1A2 gene exhibited a moderate effect size for lithium response in bipolar disorder in patients of European ancestry. SNPs associated with drug-induced organ injury were rare but exhibited the largest effect sizes, consistent with the published literature. CONCLUSION: This work demonstrates that an in silico bioinformatics-based approach using integrative analysis of a diversity of molecular and morphological data types can discover pharmacoepigenomic variants that are suitable candidates for further validation in cell lines, animal models and human clinical trials.

AIM: A regulatory network in the human brain mediating lithium response in bipolar patients was revealed by analysis of functional SNPs from genome-wide association studies (GWAS) and published gene association studies, followed by epigenome mapping. METHODS: An initial set of 23,312 SNPs in linkage disequilibrium with lead SNPs, and sub-threshold GWAS SNPs rescued by pathway analysis, were studied in the same populations. These were assessed using our workflow and annotation by the epigenome roadmap consortium. RESULTS: Twenty-seven percent of 802 SNPs that were associated with lithium response (13 published studies gene association studies and two GWAS) were shared in common with 1281 SNPs from 18 GWAS examining psychiatric disorders and adverse events associated with lithium treatment. Nineteen SNPs were annotated as active regulatory elements such as enhancers and promoters in a tissue-specific manner. They were located within noncoding regions of ten genes: ANK3, ARNTL, CACNA1C, CACNG2, CDKN1A, CREB1, GRIA2, GSK3B, NR1D1 and SLC1A2. Following gene set enrichment and pathway analysis, these genes were found to be significantly associated (p = 10(-27); Fisher exact test) with an AMPA2 glutamate receptor network in human brain. Our workflow results showed concordance with annotation of regulatory elements from the epigenome roadmap. Analysis of cognate mRNA and enhancer RNA exhibited patterns consistent with an integrated pathway in human brain. CONCLUSION: This pharmacoepigenomic regulatory pathway is located in the same brain regions that exhibit tissue volume loss in bipolar disorder. Although in silico analysis requires biological validation, the approach provides value for identification of candidate variants that may be used in pharmacogenomic testing to identify bipolar patients likely to respond to lithium.

BACKGROUND: Cell lines have been widely used in biomedical research. The community-based Cell Line Ontology (CLO) is a member of the OBO Foundry library that covers the domain of cell lines. Since its publication two years ago, significant updates have been made, including new groups joining the CLO consortium, new cell line cells, upper level alignment with the Cell Ontology (CL) and the Ontology for Biomedical Investigation, and logical extensions. CONSTRUCTION AND CONTENT: Collaboration among the CLO, CL, and OBI has established consensus definitions of cell line-specific terms such as 'cell line', 'cell line cell', 'cell line culturing', and 'mortal' vs. 'immortal cell line cell'. A cell line is a genetically stable cultured cell population that contains individual cell line cells. The hierarchical structure of the CLO is built based on the hierarchy of the in vivo cell types defined in CL and tissue types (from which cell line cells are derived) defined in the UBERON cross-species anatomy ontology. The new hierarchical structure makes it easier to browse, query, and perform automated classification. We have recently added classes representing more than 2,000 cell line cells from the RIKEN BRC Cell Bank to CLO. Overall, the CLO now contains ~38,000 classes of specific cell line cells derived from over 200 in vivo cell types from various organisms. UTILITY AND DISCUSSION: The CLO has been applied to different biomedical research studies. Example case studies include annotation and analysis of EBI ArrayExpress data, bioassays, and host-vaccine/pathogen interaction. CLO's utility goes beyond a catalogue of cell line types. The alignment of the CLO with related ontologies combined with the use of ontological reasoners will support sophisticated inferencing to advance translational informatics development.

tranSMART is an emerging global open source public private partnership community developing a comprehensive informatics-based analysis and data-sharing cloud platform for clinical and translational research. The tranSMART consortium includes pharmaceutical and other companies, not-for-profits, academic entities, patient advocacy groups, and government stakeholders. The tranSMART value proposition relies on the concept that the global community of users, developers, and stakeholders are the best source of innovation for applications and for useful data. Continued development and use of the tranSMART platform will create a means to enable "pre-competitive" data sharing broadly, saving money and, potentially accelerating research translation to cures. Significant transformative effects of tranSMART includes 1) allowing for all its user community to benefit from experts globally, 2) capturing the best of innovation in analytic tools, 3) a growing 'big data' resource, 4) convergent standards, and 5) new informatics-enabled translational science in the pharma, academic, and not-for-profit sectors.

Vaccine adverse events (VAEs) are adverse bodily changes occurring after vaccination. Understanding the adverse event (AE) profiles is a crucial step to identify serious AEs. Two different types of seasonal influenza vaccines have been used on the market: trivalent (killed) inactivated influenza vaccine (TIV) and trivalent live attenuated influenza vaccine (LAIV). Different adverse event profiles induced by these two groups of seasonal influenza vaccines were studied based on the data drawn from the CDC Vaccine Adverse Event Report System (VAERS). Extracted from VAERS were 37,621 AE reports for four TIVs (Afluria, Fluarix, Fluvirin, and Fluzone) and 3,707 AE reports for the only LAIV (FluMist). The AE report data were analyzed by a novel combinatorial, ontology-based detection of AE method (CODAE). CODAE detects AEs using Proportional Reporting Ratio (PRR), Chi-square significance test, and base level filtration, and groups identified AEs by ontology-based hierarchical classification. In total, 48 TIV-enriched and 68 LAIV-enriched AEs were identified (PRR>2, Chi-square score >4, and the number of cases >0.2% of total reports). These AE terms were classified using the Ontology of Adverse Events (OAE), MedDRA, and SNOMED-CT. The OAE method provided better classification results than the two other methods. Thirteen out of 48 TIV-enriched AEs were related to neurological and muscular processing such as paralysis, movement disorders, and muscular weakness. In contrast, 15 out of 68 LAIV-enriched AEs were associated with inflammatory response and respiratory system disorders. There were evidences of two severe adverse events (Guillain-Barre Syndrome and paralysis) present in TIV. Although these severe adverse events were at low incidence rate, they were found to be more significantly enriched in TIV-vaccinated patients than LAIV-vaccinated patients. Therefore, our novel combinatorial bioinformatics analysis discovered that LAIV had lower chance of inducing these two severe adverse events than TIV. In addition, our meta-analysis found that all previously reported positive correlation between GBS and influenza vaccine immunization were based on trivalent influenza vaccines instead of monovalent influenza vaccines.

BACKGROUND: A critical aspect of clinical and translational science (CTS) is interdisciplinary and collaborative research, which increasingly requires a wide range of computational and human resources. However, few studies have systematically analyzed such resource needs of CTS researchers. OBJECTIVE: To improve our understanding of CTS researchers' needs for computational and human resources in order to build useful and useable supporting informatics tools. METHODS: We conducted semistructured interviews of 30 CTS researchers from the University of Michigan, followed by qualitative analysis of the interview transcripts. RESULTS: The analysis identified three recurring themes: the need for the federation of information, the need to address information overload, and the need to humanize computing, including strong and well-informed views about the use of social networking tools for research collaboration. These findings helped us to narrow down the available design choices for assisting CTS researchers, and helped to identify potential deficiencies of well-known theoretical frameworks used to guide our study, with suggestions for future remedies. CONCLUSIONS: The user needs identified through the study, along with concrete design suggestions, provided key design, methodological, and theoretical insights, which are being used to guide the design and development of a CTS resource portal. The results and interview instrument should be useful to other institutions with Clinical and Translational Science Awards that face similar challenges related to helping CTS researchers make more effective use of computational and human resources.

The National Center for Integrative and Biomedical Informatics (NCIBI) is one of the eight NCBCs. NCIBI supports information access and data analysis for biomedical researchers, enabling them to build computational and knowledge models of biological systems to address the Driving Biological Problems (DBPs). The NCIBI DBPs have included prostate cancer progression, organ-specific complications of type 1 and 2 diabetes, bipolar disorder, and metabolic analysis of obesity syndrome. Collaborating with these and other partners, NCIBI has developed a series of software tools for exploratory analysis, concept visualization, and literature searches, as well as core database and web services resources. Many of our training and outreach initiatives have been in collaboration with the Research Centers at Minority Institutions (RCMI), integrating NCIBI and RCMI faculty and students, culminating each year in an annual workshop. Our future directions include focusing on the TranSMART data sharing and analysis initiative.

BACKGROUND: Metabolite profiles can be used for identifying molecular signatures and mechanisms underlying diseases since they reflect the outcome of complex upstream genomic, transcriptomic, proteomic and environmental events. The scarcity of publicly accessible large scale metabolome datasets related to human disease has been a major obstacle for assessing the potential of metabolites as biomarkers as well as understanding the molecular events underlying disease-related metabolic changes. The availability of metabolite and gene expression profiles for the NCI-60 cell lines offers the possibility of identifying significant metabolome and transcriptome features and discovering unique molecular processes related to different cancer types. METHODS: We utilized a combination of analytical methods in the R statistical package to evaluate metabolic features associated with cancer cell lines from different tissue origins, identify metabolite-gene correlations and detect outliers cell lines based on metabolome and transcriptome data. Statistical analysis results are integrated with metabolic pathway annotations as well as COSMIC and Tumorscape databases to explore associated molecular mechanisms. RESULTS: Our analysis reveals that although the NCI-60 metabolome dataset is quite noisy comparing with microarray-based transcriptome data, it does contain tissue origin specific signatures. We also identified biologically meaningful gene-metabolite associations. Most remarkably, several abnormal gene-metabolite relationships identified by our approach can be directly linked to known gene mutations and copy number variations in the corresponding cell lines. CONCLUSIONS: Our results suggest that integrative metabolome and transcriptome analysis is a powerful method for understanding molecular machinery underlying various pathophysiological processes. We expect the availability of large scale metabolome data in the coming years will significantly promote the discovery of novel biomarkers, which will in turn improve the understanding of molecular mechanism underlying diseases.

The biomedical research community relies on a diverse set of resources, both within their own institutions and at other research centers. In addition, an increasing number of shared electronic resources have been developed. Without effective means to locate and query these resources, it is challenging, if not impossible, for investigators to be aware of the myriad resources available, or to effectively perform resource discovery when the need arises. In this paper, we describe the development and use of the Biomedical Resource Ontology (BRO) to enable semantic annotation and discovery of biomedical resources. We also describe the Resource Discovery System (RDS) which is a federated, inter-institutional pilot project that uses the BRO to facilitate resource discovery on the Internet. Through the RDS framework and its associated Biositemaps infrastructure, the BRO facilitates semantic search and discovery of biomedical resources, breaking down barriers and streamlining scientific research that will improve human health.

Significant interest exists in establishing synergistic research in bioinformatics, systems biology and intelligent computing. Supported by the United States National Science Foundation (NSF), International Society of Intelligent Biological Medicine (http://www.ISIBM.org), International Journal of Computational Biology and Drug Design (IJCBDD) and International Journal of Functional Informatics and Personalized Medicine, the ISIBM International Joint Conferences on Bioinformatics, Systems Biology and Intelligent Computing (ISIBM IJCBS 2009) attracted more than 300 papers and 400 researchers and medical doctors world-wide. It was the only inter/multidisciplinary conference aimed to promote synergistic research and education in bioinformatics, systems biology and intelligent computing. The conference committee was very grateful for the valuable advice and suggestions from honorary chairs, steering committee members and scientific leaders including Dr. Michael S. Waterman (USC, Member of United States National Academy of Sciences), Dr. Chih-Ming Ho (UCLA, Member of United States National Academy of Engineering and Academician of Academia Sinica), Dr. Wing H. Wong (Stanford, Member of United States National Academy of Sciences), Dr. Ruzena Bajcsy (UC Berkeley, Member of United States National Academy of Engineering and Member of United States Institute of Medicine of the National Academies), Dr. Mary Qu Yang (United States National Institutes of Health and Oak Ridge, DOE), Dr. Andrzej Niemierko (Harvard), Dr. A. Keith Dunker (Indiana), Dr. Brian D. Athey (Michigan), Dr. Weida Tong (FDA, United States Department of Health and Human Services), Dr. Cathy H. Wu (Georgetown), Dr. Dong Xu (Missouri), Drs. Arif Ghafoor and Okan K Ersoy (Purdue), Dr. Mark Borodovsky (Georgia Tech, President of ISIBM), Dr. Hamid R. Arabnia (UGA, Vice-President of ISIBM), and other scientific leaders. The committee presented the 2009 ISIBM Outstanding Achievement Awards to Dr. Joydeep Ghosh (UT Austin), Dr. Aidong Zhang (Buffalo) and Dr. Zhi-Hua Zhou (Nanjing) for their significant contributions to the field of intelligent biological medicine.

SUMMARY: GSearcher provides a highly interactive user experience in navigating attribute data associated with large and complex biological networks. The user may either perform a quick search using keywords, phrases or regular expressions, or build a complex query with a group of filters for efficient and flexible exploration of large datasets. AVAILABILITY: http://brainarray.mbni.med.umich.edu/gsearcher/.

BACKGROUND: MicroRNAs (miRNAs) are endogenous small RNAs that modulate gene expression at the post-transcriptional level by binding complementary sites in the 3'-UTR. In a recent genome-wide study reporting a new miRNA target class (miBridge), we identified and validated interactions between 5'-UTRs and miRNAs. Separately, upstream AUGs (uAUGs) in 5'-UTRs are known to regulate genes translationally without affecting mRNA levels, one of the mechanisms for miRNA-mediated repression. RESULTS: Using sequence data from whole-genome cDNA alignments we identified 1418 uAUG sequences on the 5'-UTR that specifically interact with 3'-ends of conserved miRNAs. We computationally identified miRNAs that can target six genes through their uAUGs that were previously reported to suppress translation. We extended this meta-analysis by confirming expression of these miRNAs in cell-lines used in the uAUG studies. Similarly, seven members of the KLF family of genes containing uAUGs were computationally identified as interacting with several miRNAs. Using KLF9 as an example (whose protein expression is limited to brain tissue despite the mRNA being expressed ubiquitously), we show computationally that miRNAs expressed only in HeLa cells and not in neuroblastoma (N2A) cells can bind the uAUGs responsible for translation inhibition. Our computed results demonstrate that tissue- or cell-line specific repression of protein translation by uAUGs can be explained by the presence or absence of miRNAs that target these uAUG sequences. We propose that these uAUGs represent a subset of miRNA interaction sites on 5'-UTRs in miBridge, whereby a miRNA binding a uAUG hinders the progression of ribosome scanning the mRNA before it reaches the open reading frame (ORF). CONCLUSIONS: While both miRNAs and uAUGs are separately known to down-regulate protein expression, we show that they may be functionally related by identifying potential interactions through a sequence-specific binding mechanism. Using prior experimental evidence that shows uAUG effects on translation repression together with miRNA expression data specific to cell lines, we demonstrate through computational analysis that cell-specific down-regulation of protein expression (while maintaining mRNA levels) correlates well with the simultaneous presence of miRNA and target uAUG sequences in one cell type and not others, suggesting tissue-specific translation repression by miRNAs through uAUGs.

SUMMARY: Metscape is a plug-in for Cytoscape, used to visualize and interpret metabolomic data in the context of human metabolic networks. We have developed a metabolite database by extracting and integrating information from several public sources. By querying this database, Metscape allows users to trace the connections between metabolites and genes, visualize compound networks and display compound structures as well as information for reactions, enzymes, genes and pathways. Applying the pathway filter, users can create subnetworks that consist of compounds and reactions from a given pathway. Metscape allows users to upload experimental data, and visualize and explore compound networks over time, or experimental conditions. Color and size of the nodes are used to visualize these dynamic changes. Metscape can display the entire metabolic network or any of the pathway-specific networks that exist in the database. AVAILABILITY: Metscape can be installed from within Cytoscape 2.6.x under 'Network and Attribute I/O' category. For more information, please visit http://metscape.ncibi.org/tryplugin.html.

Clinical and translational research increasingly requires computation. Projects may involve multiple computationally oriented groups including information technology (IT) professionals, computer scientists, and biomedical informaticians. However, many biomedical researchers are not aware of the distinctions among these complementary groups, leading to confusion, delays, and suboptimal results. Although written from the perspective of Clinical and Translational Science Award (CTSA) programs within academic medical centers, this article addresses issues that extend beyond clinical and translational research. The authors describe the complementary but distinct roles of operational IT, research IT, computer science, and biomedical informatics using a clinical data warehouse as a running example. In general, IT professionals focus on technology. The authors distinguish between two types of IT groups within academic medical centers: central or administrative IT (supporting the administrative computing needs of large organizations) and research IT (supporting the computing needs of researchers). Computer scientists focus on general issues of computation such as designing faster computers or more efficient algorithms, rather than specific applications. In contrast, informaticians are concerned with data, information, and knowledge. Biomedical informaticians draw on a variety of tools, including but not limited to computers, to solve information problems in health care and biomedicine. The paper concludes with recommendations regarding administrative structures that can help to maximize the benefit of computation to biomedical research within academic health centers.

The advent of high-throughput next generation sequencing technologies have fostered enormous potential applications of supercomputing techniques in genome sequencing, epi-genetics, metagenomics, personalized medicine, discovery of non-coding RNAs and protein-binding sites. To this end, the 2008 International Conference on Bioinformatics and Computational Biology (Biocomp) - 2008 World Congress on Computer Science, Computer Engineering and Applied Computing (Worldcomp) was designed to promote synergistic inter/multidisciplinary research and education in response to the current research trends and advances. The conference attracted more than two thousand scientists, medical doctors, engineers, professors and students gathered at Las Vegas, Nevada, USA during July 14-17 and received great success. Supported by International Society of Intelligent Biological Medicine (ISIBM), International Journal of Computational Biology and Drug Design (IJCBDD), International Journal of Functional Informatics and Personalized Medicine (IJFIPM) and the leading research laboratories from Harvard, M.I.T., Purdue, UIUC, UCLA, Georgia Tech, UT Austin, U. of Minnesota, U. of Iowa etc, the conference received thousands of research papers. Each submitted paper was reviewed by at least three reviewers and accepted papers were required to satisfy reviewers' comments. Finally, the review board and the committee decided to select only 19 high-quality research papers for inclusion in this supplement to BMC Genomics based on the peer reviews only. The conference committee was very grateful for the Plenary Keynote Lectures given by: Dr. Brian D. Athey (University of Michigan Medical School), Dr. Vladimir N. Uversky (Indiana University School of Medicine), Dr. David A. Patterson (Member of United States National Academy of Sciences and National Academy of Engineering, University of California at Berkeley) and Anousheh Ansari (Prodea Systems, Space Ambassador). The theme of the conference to promote synergistic research and education has been achieved successfully.

MicroRNAs (miRNAs) are known to post-transcriptionally regulate target mRNAs through the 3'-UTR, which interacts mainly with the 5'-end of miRNA in animals. Here we identify many endogenous motifs within human 5'-UTRs specific to the 3'-ends of miRNAs. The 3'-end of conserved miRNAs in particular has significant interaction sites in the human-enriched, less conserved 5'-UTR miRNA motifs, while human-specific miRNAs have significant interaction sites only in the conserved 5'-UTR motifs, implying both miRNA and 5'-UTR are actively evolving in response to each other. Additionally, many miRNAs with their 3'-end interaction sites in the 5'-UTRs turn out to simultaneously contain 5'-end interaction sites in the 3'-UTRs. Based on these findings we demonstrate combinatory interactions between a single miRNA and both end regions of an mRNA using model systems. We further show that genes exhibiting large-scale protein changes due to miRNA overexpression or deletion contain both UTR interaction sites predicted. We provide the predicted targets of this new miRNA target class, miBridge, as an efficient way to screen potential targets, especially for nonconserved miRNAs, since the target search space is reduced by an order of magnitude compared with the 3'-UTR alone. Efficacy is confirmed by showing SEC24D regulation with hsa-miR-605, a miRNA identified only in primate, opening the door to the study of nonconserved miRNAs. Finally, miRNAs (and associated proteins) involved in this new targeting class may prevent 40S ribosome scanning through the 5'-UTR and keep it from reaching the start-codon, preventing 60S association.

Molecular dynamics simulations of nano-therapeutics as a final product and of all intermediates in the process of generating a multi-functional nano-therapeutic based on a poly(amidoamine) (PAMAM) dendrimer were performed along with chemical analyses of each of them. The actual structures of the dendrimers were predicted, based on potentiometric titration, gel permeation chromatography, and NMR. The chemical analyses determined the numbers of functional molecules, based on the actual structure of the dendrimer. Molecular dynamics simulations calculated the configurations of the intermediates and the radial distributions of functional molecules, based on their numbers. This interactive process between the simulation results and the chemical analyses provided a further strategy to design the next reaction steps and to gain insight into the products at each chemical reaction step.

Velos eResearch is a commercially-developed, regulatory-compliant, web-based clinical research information system from Velos Inc. Aithent Inc. is a software development services company. The University of Michigan (UM) has public/private partnerships with Velos and Aithent to collaborate on development of additional capabilities, modules, and new products to better support the needs of clinical and translational research communities. These partnerships provide UM with a mechanism for obtaining high-quality functionally comprehensive capabilities more quickly and at lower cost, while the corporate partners get a quality advisory and development partner--this benefits all parties. The UM chose to partner with Velos in part because of its commitment to interoperability. Velos is an active participant in the NCI caBIG project and is committed to caBIG compatibility. Velos already provides interoperability with other Velos sites in the CTSA context. One example of the partnership is co-development of integrated specimen management capabilities. UM spent more than a year defining business requirements and technical specifications for, and is funding development of, this capability. UM also facilitates an autonomous user community (20+ institutions, 7 CTSA awardees); the broad goal of the group is to share experiences, expertise, identify collaborative opportunities, and support one another as well as provide a source of future needs identification to Velos. Advantages and risks related to delivering informatics capabilities to an AHC research community through a public/private partnership will be presented. The UM, Velos and Aithent will discuss frameworks, agreements and other factors that have contributed to a successful partnership.

MOTIVATION: Cell lines are used extensively in biomedical research, but the nomenclature describing cell lines has not been standardized. The problems are both linguistic and experimental. Many ambiguous cell line names appear in the published literature. Users of the same cell line may refer to it in different ways, and cell lines may mutate or become contaminated without the knowledge of the user. As a first step towards rationalizing this nomenclature, we created a cell line knowledgebase (CLKB) with a well-structured collection of names and descriptive data for cell lines cultured in vitro. The objectives of this work are: (i) to assist users in extracting useful information from biomedical text and (ii) to highlight the importance of standardizing cell line names in biomedical research. This CLKB contains a broad collection of cell line names compiled from ATCC, Hyper CLDB and MeSH. In addition to names, the knowledgebase specifies relationships between cell lines. We analyze the use of cell line names in biomedical text. Issues include ambiguous names, polymorphisms in the use of names and the fact that some cell line names are also common English words. Linguistic patterns associated with the occurrence of cell line names are analyzed. Applying these patterns to find additional cell line names in the literature identifies only a small number of additional names. Annotation of microarray gene expression studies is used as a test case. The CLKB facilitates data exploration and comparison of different cell lines in support of clinical and experimental research. AVAILABILITY: The web ontology file for this cell line collection can be downloaded at http://www.stateslab.org/data/celllineOntology/cellline.zip.

The advancement of the computational biology field hinges on progress in three fundamental directions--the development of new computational algorithms, the availability of informatics resource management infrastructures and the capability of tools to interoperate and synergize. There is an explosion in algorithms and tools for computational biology, which makes it difficult for biologists to find, compare and integrate such resources. We describe a new infrastructure, iTools, for managing the query, traversal and comparison of diverse computational biology resources. Specifically, iTools stores information about three types of resources--data, software tools and web-services. The iTools design, implementation and resource meta-data content reflect the broad research, computational, applied and scientific expertise available at the seven National Centers for Biomedical Computing. iTools provides a system for classification, categorization and integration of different computational biology resources across space-and-time scales, biomedical problems, computational infrastructures and mathematical foundations. A large number of resources are already iTools-accessible to the community and this infrastructure is rapidly growing. iTools includes human and machine interfaces to its resource meta-data repository. Investigators or computer programs may utilize these interfaces to search, compare, expand, revise and mine meta-data descriptions of existent computational biology resources. We propose two ways to browse and display the iTools dynamic collection of resources. The first one is based on an ontology of computational biology resources, and the second one is derived from hyperbolic projections of manifolds or complex structures onto planar discs. iTools is an open source project both in terms of the source code development as well as its meta-data content. iTools employs a decentralized, portable, scalable and lightweight framework for long-term resource management. We demonstrate several applications of iTools as a framework for integrated bioinformatics. iTools and the complete details about its specifications, usage and interfaces are available at the iTools web page http://iTools.ccb.ucla.edu.

UNLABELLED: MiSearch is an adaptive biomedical literature search tool that ranks citations based on a statistical model for the likelihood that a user will choose to view them. Citation selections are automatically acquired during browsing and used to dynamically update a likelihood model that includes authorship, journal and PubMed indexing information. The user can optionally elect to include or exclude specific features and vary the importance of timeliness in the ranking. AVAILABILITY: http://misearch.ncibi.org. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

MicroRNAs (miRNA) are endogenous tissue-specific short RNAs that regulate gene expression. Discriminating each let-7 family member expression is especially important due to let-7's abundance and connection with development and cancer. However, short lengths (22 nt) and similarities between multiple sequences have prevented identification of individual members. Here, we present ProDeG, a computational algorithm which designs imperfectly matched sequences (previously yielding only noise levels in microarray experiments) for genome-wide microarray "signal" probes to discriminate single nucleotide differences and to improve probe qualities. Our probes for the entire let-7 family are both homogeneous and specific, verified using microarray signals from fluorescent dye-tagged oligonucleotides corresponding to the let-7 family, demonstrating the power of our algorithm. In addition, false let-7c signals from conventional perfectly-matched probes were identified in lymphoblastoid cell-line samples through comparison with our probe-set signals, raising concerns about false let-7 family signals in conventional microarray platform.

PURPOSE: From the Hippocratic Oath to the World Medical Association's Declaration of Geneva, physicians have sworn to protect patients' privacy. However, as systems move to more integrated architectures, protecting this medical data becomes more of a challenge. The increase in complexity of IT environments, the aggregation of data, and the desire of other entities to access this data, often 24 h/day x 7 day/week x 365 day/year, is putting serious strains on our ability to maintain its security. This problem cuts across all electronic record sources from patient care records to academic medical research records. APPROACH: In order to address this issue, we are rethinking the way we store, transmit, process, access, and federate patient data from clinical and research applications. Our groups at the University of Michigan are developing a system called the "Honest Broker" to help manage this problem. The Honest Broker will offload the burden of housing identifiable data elements of protected health information (PHI) (e.g., name and address) as well as manage data transfer between clinical and research systems. Lab results and other non-identifiable data will be stored in separate systems with either a research study ID or clinical ID number. This two-component architecture increases the burden on attackers who now need to compromise two systems, one of which is seriously hardened, in order to match health data with a patient's actual identity. CONCLUSIONS: While no security system is truly intrusion-proof, this architecture provides a high security choke point reducing the likelihood of a breach. By redesigning the method of integrating clinical care and research, we have enabled projects that would be cost prohibitive to conduct otherwise. The scalability of this mechanism is dependant on nature of the heterogenous nature of the clinical systems serving patients.

Multiple clinical informatics systems have been developed within separate departments of the University of Michigan Medical School. We are in the process of creating an "Honest Broker" method of safely and securely linking together data from different clinical systems for a research project studying the co-morbidity of depression and cardiovascular disease. The Michigan Clinical Research Collaboratory (MCRC) is an NIH/NHLBI Roadmap initiative funded to re-engineer the clinical research enterprise.

It has long been known that image plane holography with low-coherence illumination achieves optical sectioning of a volume object. A method is analyzed that is similar to image plane holography, but the interferometric arrangement utilizes the interference between two object-bearing beams instead of the basic object and reference beams.

A theory of optical sectioning by image plane holography is developed, emphasizing the use of broad-spectrum holographic methods to enhance the process. It is shown that a broad-spectrum source in a grating interferometer imitates the behavior of a monochromatic broad source.

The method of Fourier optics is applied to the problem of time-gated imaging through scattering media. Tb adapt the problem to this treatment, appropriate alterations are made: The continuous medium is replaced by a cascade of thin scatterers, and a spatial filtering process is substituted for the conventional gating processes. Closed-form solutions are derived.

The propagation of scalar waves through random media that provide multiple elastic scattering is considered by derivation of an expression for the angular correlation of the scattered wave amplitudes. Coherent wave transmission is shown to occur through a mechanism similar to that responsible for coherent backscattering. While the properties of the scattered wave are generally consistent with radiative-transfer theory for sufficiently small incident and scattering angles, coherent transmission provides corrections to radiative-transfer results at larger angles. The theoretical angular correlation curves are fit, by specifying the probability densities of two random variables that correspond to material parameters, to measured data of laser light scattering from various polymer microsphere suspensions.

OBJECTIVE: User data and information about anatomy education were used to guide development of a learning environment that is efficient and effective. The research question focused on how to design instructional software suitable for the educational goals of different groups of users of the Visible Human data set. The ultimate goal of the study was to provide options for students and teachers to use different anatomy learning modules corresponding to key topics, for course work and professional training. DESIGN: The research used both qualitative and quantitative methods. It was driven by the belief that good instructional design must address learning context information and pedagogic content information. The data collection emphasized measurement of users' perspectives, experience, and demands in anatomy learning. MEASUREMENT: Users' requirements elicited from 12 focus groups were combined and rated by 11 researchers. Collective data were sorted and analyzed by use of multidimensional scaling and cluster analysis. RESULTS: A set of functions and features in high demand across all groups of users was suggested by the results. However, several subgroups of users shared distinct demands. The design of the learning modules will encompass both unified core components and user-specific applications. The design templates will allow sufficient flexibility for dynamic insertion of different learning applications for different users. CONCLUSION: This study describes how users' requirements, associated with users' learning experiences, were systematically collected and analyzed and then transformed into guidelines informing the iterative design of multiple learning modules. Information about learning challenges and processes was gathered to define essential anatomy teaching strategies. A prototype instrument to design and polish the Visible Human user interface system is currently being developed using ideas and feedback from users.

With a spatial-filtering method of gating, we explore image formation through scattering media using first-arriving light. Gating times of a few femtoseconds and less are produced, and the resolution at these extremely short gating times is investigated.

Laser scanning confocal microscopy was used to determine the distribution of actin, spectrin and tubulin in whole mounts of the organ of Corti of guinea pig, monkey, rat and chinchilla. Actin, spectrin and tubulin were localized in all cell types in the auditory epithelium. No specialized cytoskeletal organization of tubulin was detected in the cytoplasmic domain of hair cells. The only specialized organization of actin and spectrin in the cytoplasmic domain was the infra-cuticular network, found exclusively in apical guinea pig outer hair cells. In contrast, the lateral wall of inner and outer hair cells contained a homogeneous distribution of label specific for actin and spectrin. The label intensity was similar in the base and the apex of the cochlea. These results indicate that the distribution of spectrin and actin in the auditory epithelium is similar to that in other epithelial cells, suggesting that actin and spectrin participate in the formation of cellular shape and possibly in docking molecules to the membrane.