Showing papers by "James B. Brown published in 2021"

CATMoS: Collaborative Acute Toxicity Modeling Suite.

Abstract: Background: Humans are exposed to tens of thousands of chemical substances that need to be assessed for their potential toxicity. Acute systemic toxicity testing serves as the basis for regulatory ...

Selective time-dependent changes in activity and cell-specific gene expression in human postmortem brain.

Abstract: As a means to understand human neuropsychiatric disorders from human brain samples, we compared the transcription patterns and histological features of postmortem brain to fresh human neocortex isolated immediately following surgical removal. Compared to a number of neuropsychiatric disease-associated postmortem transcriptomes, the fresh human brain transcriptome had an entirely unique transcriptional pattern. To understand this difference, we measured genome-wide transcription as a function of time after fresh tissue removal to mimic the postmortem interval. Within a few hours, a selective reduction in the number of neuronal activity-dependent transcripts occurred with relative preservation of housekeeping genes commonly used as a reference for RNA normalization. Gene clustering indicated a rapid reduction in neuronal gene expression with a reciprocal time-dependent increase in astroglial and microglial gene expression that continued to increase for at least 24 h after tissue resection. Predicted transcriptional changes were confirmed histologically on the same tissue demonstrating that while neurons were degenerating, glial cells underwent an outgrowth of their processes. The rapid loss of neuronal genes and reciprocal expression of glial genes highlights highly dynamic transcriptional and cellular changes that occur during the postmortem interval. Understanding these time-dependent changes in gene expression in post mortem brain samples is critical for the interpretation of research studies on human brain disorders.

3D-Scaffold: A Deep Learning Framework to Generate 3D Coordinates of Drug-like Molecules with Desired Scaffolds.

Abstract: The prerequisite of therapeutic drug design and discovery is to identify novel molecules and developing lead candidates with desired biophysical and biochemical properties. Deep generative models have demonstrated their ability to find such molecules by exploring a huge chemical space efficiently. An effective way to generate new molecules with desired target properties is by constraining the critical fucntional groups or the core scaffolds in the generation process. To this end, we developed a domain aware generative framework called 3D-Scaffold that takes 3D coordinates of the desired scaffold as an input and generates 3D coordinates of novel therapeutic candidates as an output while always preserving the desired scaffolds in generated structures. We demonstrated that our framework generates predominantly valid, unique, novel, and experimentally synthesizable molecules that have drug-like properties similar to the molecules in the training set. Using domain specific data sets, we generate covalent and noncovalent antiviral inhibitors targeting viral proteins. To measure the success of our framework in generating therapeutic candidates, generated structures were subjected to high throughput virtual screening via docking simulations, which shows favorable interaction against SARS-CoV-2 main protease (Mpro) and nonstructural protein endoribonuclease (NSP15) targets. Most importantly, our deep learning model performs well with relatively small 3D structural training data and quickly learns to generalize to new scaffolds, highlighting its potential application to other domains for generating target specific candidates.

Influence of soil heterogeneity on soybean plant development and crop yield evaluated using time-series of UAV and ground-based geophysical imagery.

Abstract: Understanding the interactions among agricultural processes, soil, and plants is necessary for optimizing crop yield and productivity. This study focuses on developing effective monitoring and analysis methodologies that estimate key soil and plant properties. These methodologies include data acquisition and processing approaches that use unmanned aerial vehicles (UAVs) and surface geophysical techniques. In particular, we applied these approaches to a soybean farm in Arkansas to characterize the soil-plant coupled spatial and temporal heterogeneity, as well as to identify key environmental factors that influence plant growth and yield. UAV-based multitemporal acquisition of high-resolution RGB (red-green-blue) imagery and direct measurements were used to monitor plant height and photosynthetic activity. We present an algorithm that efficiently exploits the high-resolution UAV images to estimate plant spatial abundance and plant vigor throughout the growing season. Such plant characterization is extremely important for the identification of anomalous areas, providing easily interpretable information that can be used to guide near-real-time farming decisions. Additionally, high-resolution multitemporal surface geophysical measurements of apparent soil electrical conductivity were used to estimate the spatial heterogeneity of soil texture. By integrating the multiscale multitype soil and plant datasets, we identified the spatiotemporal co-variance between soil properties and plant development and yield. Our novel approach for early season monitoring of plant spatial abundance identified areas of low productivity controlled by soil clay content, while temporal analysis of geophysical data showed the impact of soil moisture and irrigation practice (controlled by topography) on plant dynamics. Our study demonstrates the effective coupling of UAV data products with geophysical data to extract critical information for farm management.

The Time Machine framework: monitoring and prediction of biodiversity loss.

Abstract: Transdisciplinary solutions are needed to achieve the sustainability of ecosystem services for future generations. We propose a framework to identify the causes of ecosystem function loss and to forecast the future of ecosystem services under different climate and pollution scenarios. The framework (i) applies an artificial intelligence (AI) time-series analysis to identify relationships among environmental change, biodiversity dynamics and ecosystem functions; (ii) validates relationships between loss of biodiversity and environmental change in fabricated ecosystems; and (iii) forecasts the likely future of ecosystem services and their socioeconomic impact under different pollution and climate scenarios. We illustrate the framework by applying it to watersheds, and provide system-level approaches that enable natural capital restoration by associating multidecadal biodiversity changes to chemical pollution.

Endothelial activation is associated with intestinal epithelial injury, systemic inflammation and treatment regimen in children living with vertically acquired HIV-1 infection

Abstract: BACKGROUND Premature development of cardiovascular disease in children living with HIV-1 (CLWH) may be associated with compromised gut barrier function, microbial translocation, immune activation, systemic inflammation and endothelial activation. Biomarkers of these pathways may provide insights into pathogenesis of atherosclerotic disease in CLWH. METHODS This was a cross-sectional study of CLWH enrolled in the multicentre Early Pediatric Initiation-Canadian Child Cure Cohort (EPIC4 ) who were on antiretroviral therapy (ART) with undetectable viral load. Plasma biomarkers of intestinal epithelial injury [intestinal fatty acid binding protein-1 (IFABP)], systemic inflammation [tumour necrosis factor (TNF) and interleukin-6 (IL-6)] and endothelial activation [angiopoietin-2 (Ang2), soluble vascular endothelial growth factor-1 (sVEGFR1) and soluble endoglin (sEng)] were quantified by enzyme-linked immunosorbent assay. Correlation and factor analysis of biomarkers were used to examine associations between innate immune pathways. RESULTS Among 90 CLWH, 16% of Ang2, 15% of sVEGFR1 and 23% of sEng levels were elevated relative to healthy historic controls. Pairwise rank correlations between the three markers of endothelial activation were statistically significant (ρ = 0.69, ρ = 0.61 and ρ = 0.65, P < 0.001 for all correlations). An endothelial activation index, derived by factor analysis of the three endothelial biomarkers, was correlated with TNF (ρ = 0.47, P < 0.001), IL-6 (ρ = 0.60, P < 0.001) and intestinal fatty acid binding protein-1 (ρ = 0.67, P < 0.001). Current or past treatment with ritonavir-boosted lopinavir (LPV/r) was associated with endothelial activation (odds ratio = 5.0, 95% CI: 1.7-17, P = 0.0020). CONCLUSIONS Endothelial activation is prevalent in CLWH despite viral suppression with combination ART and is associated with intestinal epithelial injury, systemic inflammation and treatment with LPV/r.

Utility of Homologous Recombination Deficiency Biomarkers Across Cancer Types.

Abstract: PURPOSEHomologous recombination DNA repair deficiency (HRD) is associated with sensitivity to platinum and poly (ADP-ribose) polymerase inhibitors in certain cancer types, including breast, ovarian...

Figla promotes secondary follicle growth in mature mice.

Abstract: The in vitro growth (IVG) of human follicles is a potential fertility option for women for whom cryopreserved ovarian tissues cannot be transplanted due to the risk of cancer cell reintroduction; however, there is currently no established method. Furthermore, optimal IVG conditions may differ between the follicles of adult and pre-pubertal females due to molecular differences suggested by basic research. To systematically identify differences between the secondary follicles of adult and pre-pubertal females, a comparative transcriptomic study using mice was conducted herein. Among differentially expressed genes (DEGs), Figla was up-regulated in mature mice. We successfully down-regulated Figla expression in secondary follicle oocytes by a Figla siRNA microinjection, and the subsequent IVG of follicles showed that the diameter of these follicles was smaller than those of controls in mature mice, whereas no significant difference was observed in premature mice. The canonical pathways of DEGs between control and Figla-reduced secondary follicles suggest that Figla up-regulates VDR/RXR activation and down-regulates stem cell pluripotency as well as estrogen signaling. We demonstrated for the first time that folliculogenesis of the secondary follicles of premature and mature mice may be regulated by different factors, such as Figla with its possible target genes, providing insights into optimal IVG conditions for adult and pre-pubertal females, respectively.

A novel random forest approach to revealing interactions and controls on chlorophyll concentration and bacterial communities during coastal phytoplankton blooms.

Abstract: Increasing occurrence of harmful algal blooms across the land-water interface poses significant risks to coastal ecosystem structure and human health. Defining significant drivers and their interactive impacts on blooms allows for more effective analysis and identification of specific conditions supporting phytoplankton growth. A novel iterative Random Forests (iRF) machine-learning model was developed and applied to two example cases along the California coast to identify key stable interactions: (1) phytoplankton abundance in response to various drivers due to coastal conditions and land-sea nutrient fluxes, (2) microbial community structure during algal blooms. In Example 1, watershed derived nutrients were identified as the least significant interacting variable associated with Monterey Bay phytoplankton abundance. In Example 2, through iRF analysis of field-based 16S OTU bacterial community and algae datasets, we independently found stable interactions of prokaryote abundance patterns associated with phytoplankton abundance that have been previously identified in laboratory-based studies. Our study represents the first iRF application to marine algal blooms that helps to identify ocean, microbial, and terrestrial conditions that are considered dominant causal factors on bloom dynamics.

An infrared spectral biomarker accurately predicts neurodegenerative disease class in the absence of overt symptoms.

Abstract: Although some neurodegenerative diseases can be identified by behavioral characteristics relatively late in disease progression, we currently lack methods to predict who has developed disease before the onset of symptoms, when onset will occur, or the outcome of therapeutics. New biomarkers are needed. Here we describe spectral phenotyping, a new kind of biomarker that makes disease predictions based on chemical rather than biological endpoints in cells. Spectral phenotyping uses Fourier Transform Infrared (FTIR) spectromicroscopy to produce an absorbance signature as a rapid physiological indicator of disease state. FTIR spectromicroscopy has over the past been used in differential diagnoses of manifest disease. Here, we report that the unique FTIR chemical signature accurately predicts disease class in mouse with high probability in the absence of brain pathology. In human cells, the FTIR biomarker accurately predicts neurodegenerative disease class using fibroblasts as surrogate cells.

An integrated host-microbiome response to atrazine exposure mediates toxicity in Drosophila

Abstract: The gut microbiome produces vitamins, nutrients, and neurotransmitters, and helps to modulate the host immune system-and also plays a major role in the metabolism of many exogenous compounds, including drugs and chemical toxicants. However, the extent to which specific microbial species or communities modulate hazard upon exposure to chemicals remains largely opaque. Focusing on the effects of collateral dietary exposure to the widely used herbicide atrazine, we applied integrated omics and phenotypic screening to assess the role of the gut microbiome in modulating host resilience in Drosophila melanogaster. Transcriptional and metabolic responses to these compounds are sex-specific and depend strongly on the presence of the commensal microbiome. Sequencing the genomes of all abundant microbes in the fly gut revealed an enzymatic pathway responsible for atrazine detoxification unique to Acetobacter tropicalis. We find that Acetobacter tropicalis alone, in gnotobiotic animals, is sufficient to rescue increased atrazine toxicity to wild-type, conventionally reared levels. This work points toward the derivation of biotic strategies to improve host resilience to environmental chemical exposures, and illustrates the power of integrative omics to identify pathways responsible for adverse health outcomes.

The utility of homologous recombination deficiency biomarkers across cancer types

Abstract: Background Genomic alterations in BRCA1/2 and genomic scar signatures are associated with homologous recombination DNA repair deficiency (HRD) and serve as therapeutic biomarkers for platinum and PARP inhibitors in breast and ovarian cancers. However, the clinical significance of these biomarkers in other homologous recombination repair-related genes or other cancer types is not fully understood. Results We analyzed the datasets of all solid cancers from The Cancer Genome Atlas and Cancer Cell Line Encyclopedia, and found that the association between biallelic alterations in the homologous recombination pathway genes and genomic scar signatures differed greatly depending on gender and the presence of somatic TP53 mutation. Additionally, HRD cases identified by a combination of these indicators showed higher sensitivity to DNA-damaging drugs than non-HRD cases both in clinical samples and cell lines. Conclusion Our work provides novel proof of the utility of HRD analysis for all cancer types and will improve the precision and efficacy of chemotherapy selection in clinical oncology.

Designing, implementing, and testing hardware for cybersecurity

Abstract: Cybersecurity is one of the key issues facing the world today. With an ever-increasing number of devices connected across the internet, the need to secure all these different devices against potential attackers is an endless effort. This thesis is focussed on the most promising new developments in the hardware aspect of this battle for security. The first section of the thesis looks at what is the current state of the art when it comes to hardware security primitives, with a focus on random number generators and Physically Unclonable Functions (PUF). The strengths and weakness of the current implementations of these systems are analysed so that the areas which are most in need of improvement can be highlighted. The second major section of this thesis is looking to improve how random numbers are generated, which is essential for many current security systems. True random number generators have been presented as a potential solution to this problem but improvements in output bit rate, power consumption, and design complexity must be made. In this work we present a novel and experimentally verified true random number generator that exclusively uses conventional CMOS technology as well as offering key improvements over previous designs in complexity, output bit rate, and power consumption. It uses the inherent randomness of telegraph noise in the channel current of a single CMOS transistor as an entropy source. For the first time, multi-level and abnormal telegraph noise can be utilised, which greatly reduces device selectivity and offers much greater bit rates. The design is verified using a breadboard and FPGA proof of concept circuit and passes all 15 of the NIST randomness tests without any need for post-processing of the generated bitstream. The design also shows resilience against machine learning attacks performed by an LSTM neural network. The third major section describes the development of a novel PUF concept, which offers a new approach to authentication, allowing low power devices to be included in existing networks without compromising overall security. The new PUF concept introduces time dependence to vastly increase the efficiency of entropy source usage, when compared with a traditional PUF. This new PUF also introduces a probability-based model which greatly reduces the required server memory for Challenge Response Pair (CRP) storage when large numbers of CRPs are used. The concept is verified experimentally on nano-scale CMOS technology as well as through simulation and a proof-of-concept circuit. These combined benefits bring the PUF concept much closer to being a viable solution for widespread cybersecurity applications.

Combination of gene set signatures correlates with response to nivolumab in platinum-resistant ovarian cancer

Abstract: Based on our previous phase II clinical trial of anti-programmed death-1 (PD-1) antibody nivolumab for platinum-resistant ovarian cancer (n = 19, UMIN000005714), we aimed to identify the biomarkers predictive of response. Tumor gene expression was evaluated by proliferative, mesenchymal, differentiated, and immunoreactive gene signatures derived from high-grade serous carcinomas and a signature established prior for ovarian clear cell carcinoma. Resulting signature scores were statistically assessed with both univariate and multivariate approaches for correlation to clinical response. Analyses were performed to identify pathways differentially expressed by either the complete response (CR) or progressive disease (PD) patient groups. The clear cell gene signature was scored significantly higher in the CR group, and the proliferative gene signature had significantly higher scores in the PD group where nivolumab was not effective (respective p values 0.005 and 0.026). Combinations of gene signatures improved correlation with response, where a visual projection of immunoreactive, proliferative, and clear cell signatures differentiated clinical response. An applicable clinical response prediction formula was derived. Ovarian cancer-specific gene signatures and related pathway scores provide a robust preliminary indicator for ovarian cancer patients prior to anti-PD-1 therapy decisions.

Spatial Graph Attention and Curiosity-driven Policy for Antiviral Drug Discovery

Abstract: Author(s): Wu, Yulun; Choma, Nicholas; Chen, Andrew; Cashman, Mikaela; Prates, Erica T; Shah, Manesh; Vergara, Veronica G Melesse; Clyde, Austin; Brettin, Thomas S; Jong, Wibe A de; Kumar, Neeraj; Head, Martha S; Stevens, Rick L; Nugent, Peter; Jacobson, Daniel A; Brown, James B | Abstract: We developed Distilled Graph Attention Policy Networks (DGAPNs), a curiosity-driven reinforcement learning model to generate novel graph-structured chemical representations that optimize user-defined objectives by efficiently navigating a physically constrained domain. The framework is examined on the task of generating molecules that are designed to bind, noncovalently, to functional sites of SARS-CoV-2 proteins. We present a spatial Graph Attention Network (sGAT) that leverages self-attention over both node and edge attributes as well as encoding spatial structure -- this capability is of considerable interest in areas such as molecular and synthetic biology and drug discovery. An attentional policy network is then introduced to learn decision rules for a dynamic, fragment-based chemical environment, and state-of-the-art policy gradient techniques are employed to train the network with enhanced stability. Exploration is efficiently encouraged by incorporating innovation reward bonuses learned and proposed by random network distillation. In experiments, our framework achieved outstanding results compared to state-of-the-art algorithms, while increasing the diversity of proposed molecules and reducing the complexity of paths to chemical synthesis.

Learning from learning machines: a new generation of AI technology to meet the needs of science

Abstract: We outline emerging opportunities and challenges to enhance the utility of AI for scientific discovery. The distinct goals of AI for industry versus the goals of AI for science create tension between identifying patterns in data versus discovering patterns in the world from data. If we address the fundamental challenges associated with "bridging the gap" between domain-driven scientific models and data-driven AI learning machines, then we expect that these AI models can transform hypothesis generation, scientific discovery, and the scientific process itself.

Enhanced neoplasia detection in chronic ulcerative colitis: the ENDCaP-C diagnostic accuracy study

Abstract: Background Chronic ulcerative colitis is a large bowel inflammatory condition associated with increased colorectal cancer risk over time, resulting in 1000 colectomies per year in the UK. Despite intensive colonoscopic surveillance, 50% of cases progress to invasive cancer before detection. Detecting early (precancer) molecular changes by analysing biopsies from routine colonoscopy should increase neoplasia detection. Objectives To establish a deoxyribonucleic acid (DNA) marker panel associated with early neoplastic changes in ulcerative colitis patients. To develop the DNA methylation test for high-throughput analysis within the NHS. To prospectively evaluate the test within the existing colonoscopy surveillance programme. Design Module 1 analysed 569 stored biopsies from neoplastic and non-neoplastic sites/patients using pyrosequencing for 11 genes that were previously reported to have altered promoter methylation associated with colitis-associated neoplasia. Classifiers were constructed to predict neoplasia based on gene combinations. Module 2 translated analysis to a NHS laboratory, assessing next-generation sequencing to increase speed and reduce cost. Module 3 applied the molecular classifiers within a prospective diagnostic accuracy study, in the existing ulcerative colitis surveillance programme. Comparisons were made between baseline and reference colonoscopies undertaken in a stratified patient sample 6–12 months later. Setting Thirty-one UK hospitals. Participants Patients with chronic ulcerative colitis, either for at least 10 years and extensive disease, or with primary sclerosing cholangitis. Interventions An optimised DNA methylation classifier tested on routine mucosal biopsies taken during colonoscopy. Main outcome Identifying ulcerative colitis patients with neoplasia. Results Module 1 selected five genes with specificity for neoplasia. The optimism-adjusted area under the receiver operating characteristic curve for neoplasia was 0.83 (95% confidence interval 0.79 to 0.88). Precancerous neoplasia showed a higher area under the receiver operating characteristic curve of 0.88 (95% confidence interval 0.84 to 0.92). Background mucosa had poorer discrimination (optimism-adjusted area under the receiver operating characteristic curve was 0.68, 95% confidence interval 0.62 to 0.73). Module 2 was unable to develop a robust next-generation sequencing assay because of the low amplification rates across all genes. In module 3, 818 patients underwent a baseline colonoscopy. The methylation assay (testing non-neoplastic mucosa) was compared with pathology assessments for neoplasia and showed a diagnostic odds ratio of 2.37 (95% confidence interval 1.46 to 3.82; p = 0.0002). The probability of dysplasia increased from 11.1% before testing to 17.7% after testing (95% confidence interval 13.0% to 23.2%), with a positive methylation result suggesting added value in neoplasia detection. To determine added value above colonoscopy alone, a second (reference) colonoscopy was performed in 193 patients without neoplasia. Although the test showed an increased number of patients with neoplasia associated with primary methylation changes, this failed to reach statistical significance (diagnostic odds ratio 3.93; 95% confidence interval 0.82 to 24.75; p = 0.09). Limitations Since the inception of ENDCaP-C, technology has advanced to allow whole-genome or methylome testing to be performed. Conclusions Methylation testing for chronic ulcerative colitis patients cannot be recommended based on this study. However, following up this cohort will reveal further neoplastic changes, indicating whether or not this test may be identifying a population at risk of future neoplasia and informing future surveillance programmes. Trial registration Current Controlled Trials ISRCTN81826545. Funding This project was funded by the Efficacy and Mechanism Evaluation programme, a Medical Research Council and National Institute for Health Research (NIHR) partnership, and will be published in full in Efficacy and Mechanism Evaluation; Vol. 8, No. 1. See the NIHR Journals Library website for further project information.

Tumor genomic subtypes orthogonal to mutation burden predict the efficacy of immune checkpoint therapy

Abstract: Background: In cancer therapy, precise tumor-agnostic biomarkers that predict response to immune checkpoint inhibitors (ICIs) are needed. To explain treatment response differences among tumor types, the application of mutational signatures, patterns of genomic alterations that reflect differences in distinct underlying carcinogenic processes, holds promise but has not been extensively integrated into prediction methodologies. Methods: Based on mutational signature analysis, we developed a stratification for all solid tumors in The Cancer Genome Atlas (TCGA). Then, we developed the Tumor Genomic Subtype Analyzer (TGSA) to classify tumors submitted to whole-exome sequencing. Using existing data from 938 pan-cancer ICI-treated cases with outcomes, we evaluated the subtype-response predictive performance. Results: Systematic analysis on TCGA samples identified eight tumor genomic subtypes, which were characterized by features represented by smoking exposure, ultraviolet light exposure, APOBEC enzyme activity, POLE mutation, mismatch repair deficiency, homologous recombination deficiency, genomic stability, and aging. The former five subtypes were presumed to form an immune-responsive group acting as candidates for ICI therapy because of their high expression of immune-related genes and enrichment in cancer types with FDA approval for ICI monotherapy. In the validation cohort, the samples assigned by TGSA to the immune-reactive subtypes were significantly related to ICI response independent of cancer type and high TMB status. Conclusion: Mutational signature-based tumor subtyping can serve as a tumor-agnostic biomarker for ICI response prediction. The results indicate that the mutational process underlying carcinogenesis affects tumor immunogenicity, and thus sensitivity to ICI.