Large Datasets Research Articles

Estimating global trends of air pollution, air pollution-attributable disease burdens, and CO2 emissions in 13,000 cities using large geospatial datasets

Estimating global trends of air pollution, air pollution-attributable disease burdens, and CO2 emissions in 13,000 cities using large geospatial datasets

Accelerated dimensionality reduction of single-cell RNA sequencing data with fastglmpca.

Motivated by theoretical and practical issues that arise when applying Principal Component Analysis (PCA) to count data, Townes et al introduced "Poisson GLM-PCA", a variation of PCA adapted to count data, as a tool for dimensionality reduction of single-cell RNA sequencing (scRNA-seq) data. However, fitting GLM-PCA is computationally challenging. Here we study this problem, and show that a simple algorithm, which we call "Alternating Poisson Regression" (APR), produces better quality fits, and in less time, than existing algorithms. APR is also memory-efficient, and lends itself to parallel implementation on multi-core processors, both of which are helpful for handling large scRNA-seq data sets. We illustrate the benefits of this approach in three publicly available scRNA-seq data sets. The new algorithms are implemented in an R package, fastglmpca. The fastglmpca R package is released on CRAN for Windows, macOS and Linux, and the source code is available at github.com/stephenslab/fastglmpca under the open source GPL-3 license. Scripts to reproduce the results in this paper are also available in the GitHub repository and on Zenodo. Supplementary data are available at Bioinformatics online.

Semisupervised Learning to Boost hERG, Nav1.5, and Cav1.2 Cardiac Ion Channel Toxicity Prediction by Mining a Large Unlabeled Small Molecule Data Set.

Predicting drug toxicity is a critical aspect of ensuring patient safety during the drug design process. Although conventional machine learning techniques have shown some success in this field, the scarcity of annotated toxicity data poses a significant challenge in enhancing models' performance. In this study, we explore the potential of leveraging large unlabeled small molecule data sets using semisupervised learning to improve drug cardiotoxicity predictive performance across three cardiac ion channel targets: the voltage-gated potassium channel (hERG), the voltage-gated sodium channel (Nav1.5), and the voltage-gated calcium channel (Cav1.2). We extensively mined the ChEMBL database, comprising approximately 2 million small molecules, and then employed semisupervised learning to construct robust classification models for this purpose. We achieved a performance boost on highly diverse (i.e., structurally dissimilar) test data sets across all three targets. Using our built models, we screened the whole ChEMBL database and a large set of FDA-approved drugs, identifying several compounds with potential cardiac ion channel activity. To ensure broad accessibility and usability for both technical and nontechnical users, we developed a cross-platform graphical user interface that allows users to make predictions and gain insights into the cardiotoxicity of drugs and other small molecules. The software is made available as open source under the permissive MIT license at https://github.com/issararab/CToxPred2.

A python library for the fast and scalable computation of biologically meaningful individual specific networks

Individual Specific Networks (ISNs) are a tool used in computational biology to infer Individual Specific relationships between biological entities from omics data. ISNs provide insights into how the interactions among these entities affect their respective functions. To address the scarcity of solutions for efficiently computing ISNs on large biological datasets, we present ISN-tractor, a data-agnostic, highly optimized Python library to build and analyse ISNs. ISN-tractor demonstrates superior scalability and efficiency in generating Individual Specific Networks (ISNs) when compared to existing methods such as LionessR, both in terms of time and memory usage, allowing ISNs to be used on large datasets. We show how ISN-tractor can be applied to real-life datasets, including The Cancer Genome Atlas (TCGA) and HapMap, showcasing its versatility. ISN-tractor can be used to build ISNs from various -omics data types, including transcriptomics, proteomics, and genotype arrays, and can detect distinct patterns of gene interactions within and across cancer types. We also show how Filtration Curves provided valuable insights into ISN characteristics, revealing topological distinctions among individuals with different clinical outcomes. Additionally, ISN-tractor can effectively cluster populations based on genetic relationships, as demonstrated with Principal Component Analysis on HapMap data.

Routine Data Analysis of HIV Pre-Exposure Prophylaxis Use and Rates of Sexually Transmitted Infections Since Coverage of HIV Pre-Exposure Prophylaxis by the Statutory Health Insurance in Germany.

Human immunodeficiency virus pre-exposure prophylaxis (PrEP) is considered as an effective protection against a human immunodeficiency virus (HIV) infection. However, it is still unclear, how PrEP use is associated with the incidence of sexually transmitted infections (STI) other than HIV. PrEP became reimbursable in Germany from September 1, 2019 for persons with statutory health insurance (SHI). With the EvE-PrEP study, the Federal Ministry of Health commissioned the evaluation of the effects of the new reimbursement situation in Germany. In the presented module of the EvE-PrEP study, routine data from three large German SHI funds were analyzed in anonymized form for the period January 1, 2019 to March 31, 2020. Data were analyzed regarding: Characteristics and adherence of PrEP users, treatment success of PrEP, and changes in STI incidence rates before and since PrEP use. The cooperating SHI funds collectively covered about 52% of the overall population in Germany in 2019. A total of 7102 persons with PrEP use were included into the analysis. These were predominantly male (99%), on average 37.4years old and a high proportion of persons lived in large cities. The average quotient of PrEP daily defined doses and assumed days on PrEP was 87%. The average STI rates normalized per 100 person-years at individual level pre PrEP did not statistically significant differ compared to since PrEP (chlamydia: 17.5 vs. 17.6, gonococcal infection: 29.1 vs. 30.7, and syphilis: 14.6 vs.13.6). A large data set was used to evaluate the introduction of PrEP as a SHI benefit in Germany. A potentially suspected increase in bacterial STI incidence rates was not found. A rather high average adherence rate was observed. The very high proportion of men and people from the largest German cities among PrEP users is striking. These results could indicate barriers to PrEP access for people at risk of HIV, especially if they are women or people living in less urban areas.

Efficient inference of large prokaryotic pangenomes with PanTA

Pangenome inference is an indispensable step in bacterial genomics, yet its scalability poses a challenge due to the rapid growth of genomic collections. This paper presents PanTA, a software package designed for constructing pangenomes of large bacterial datasets, showing unprecedented efficiency levels multiple times higher than existing tools. PanTA introduces a novel mechanism to construct the pangenome progressively without rebuilding the accumulated collection from scratch. The progressive mode is shown to consume orders of magnitude less computational resources than existing solutions in managing growing datasets. The software is open source and is publicly available at https://github.com/amromics/panta and at 10.6084/m9.figshare.23724705.

JmBIG: enhancing dynamic risk prediction and personalized medicine through joint modeling of longitudinal and survival data in big routinely collected data

We have introduced the R package jmBIG to facilitate the analysis of large healthcare datasets and the development of predictive models. This package provides a comprehensive set of tools and functions specifically designed for the joint modelling of longitudinal and survival data in the context of big data analytics. The jmBIG package offers efficient and scalable implementations of joint modelling algorithms, allowing for integrating large-scale healthcare datasets.By utilizing the capabilities of jmBIG, researchers and analysts can effectively handle the challenges associated with big healthcare data, such as high dimensionality and complex relationships between multiple outcomes.With the support of jmBIG, analysts can seamlessly fit Bayesian joint models, generate predictions, and evaluate the performance of the models. The package incorporates cutting-edge methodologies and harnesses the computational capabilities of parallel computing to accelerate the analysis of large-scale healthcare datasets significantly. In summary, jmBIG empowers researchers to gain deeper insights into disease progression and treatment response, fostering evidence-based decision-making and paving the way for personalized healthcare interventions that can positively impact patient outcomes on a larger scale.

Economic evaluation of a psychological surveillance and support programme in the UK police force

PurposePolice officers in the UK frequently experience traumatic events and work pressures, and these can have a detrimental impact on mental health. A psychological surveillance programme has been implemented and clinically evaluated. This paper reports an economic evaluation of the programme.Design/methodology/approachA decision analytic model was developed to explore programme costs, outcomes and return on investment for a notional cohort of 1,000 police personnel (officers and staff). The model parameters were obtained from a large data set of sequential psychological surveillance results. Changes in individual work productivity were derived from a workability measure included in the survey.FindingsThe modelled expected total cost incurred to screen 1,000 participants was £84,287 ($106,971). The expected net increase in work productivity for those receiving interventions was valued at £241,672 ($306,713). This result represents a return on investment of 187%. Sensitivity analyses showed that the model was robust to changes in key parameters.Research limitations/implicationsPolice personnel experiencing traumatic events may need specialist support from occupational health services or psychological therapy. Screening or surveillance can detect mental health problems, but its cost-effectiveness has been previously unknown.Practical implicationsThe value for money of providing surveillance within the police force to detect mental health problems is assessed in this research. The gains in terms of increased work are evident. In similar environmental working conditions, this could extend to emergency services globally.Social implicationsThe police are often called upon to deal with the lowest points in people’s lives. The findings have implications for police forces wishing to support officers to be able to provide the best of themselves in these situations.Originality/valueThere have been no previous studies in policing that have concluded that running a psychological surveillance programme in policing is financially viable. Although this requires initial investment, this paper models that the extra costs incurred are more than offset by improved work productivity.

Maternal exposure to ambient temperature and birth defects in Brazil: a nationwide case-control study of over 11 million newborns.

Maternal exposure to extreme ambient temperature during pregnancy has been proposed as a potential risk factor for birth defects. Comprehensive investigations on this association remain limited, particularly in low- and middle-income countries. This study aims to examine the association between ambient temperature exposure during pregnancy and the risk of birth defects in Brazil, contributing to the broader understanding of environmental influences on birth outcomes. Using a large dataset of over 11 million live birth records, we analyzed 12 categories of birth defects, encompassing a time frame from 2001 to 2018. Ambient temperature data were assigned at the municipality level. For the exposure assessment, we considered two biologically driven pregnancy stages by dividing the gestational period into two specific windows: the first trimester (from week 1 to week 12) and the second trimester (from week 13 to week 28). We employed a two-stage case-control design. In the first stage, we applied a conditional logistic regression model to estimate the odds ratio (OR) for specific birth defects and each of the five Brazilian regions (North, Northeast, Midwest, Southeast, and South). The model was adjusted for potential confounding variables, including PM2.5, relative humidity, and socioeconomic status. Temporal trends were addressed using time-stratified sampling. In the second stage, we used mixed-effects meta-analysis to pool region-specific estimates. Our analysis revealed a significant association between maternal exposure to higher ambient temperatures during the first trimester and an increased risk of specific birth defect categories, including those affecting the genital organs (OR = 1.08, 95% CI: 1.02; 1.14), digestive system (OR = 1.12, 95% CI: 1.06; 1.19); circulatory system (OR = 1.08, 95% CI: 1.01; 1.17); eyes, ears, face, and neck (OR = 1.08, 95% CI: 1.02; 1.15); benign neoplasms tumors (OR = 1.17, 95% CI: 1.03; 1.32), musculoskeletal system (OR = 1.03, 95% CI: 1.01; 1.05); and other congenital anomalies (OR = 1.22, 95% CI: 1.15; 1.29). The associations with respiratory system, nervous system, and chromosomal anomalies were null. These findings have significant implications for public health policies aimed at mitigating the impact of environmental factors on birth outcomes, both in Brazil and globally.

What teachers can do to enhance students’ well-being: Discussion

BackgroundThis special issue sheds light on how students’ well-being can be enhanced through teaching practices, emphasizing the significance of well-being in the educational context. My discussion was rooted in extensive well-being literature, which allowed me to outline key findings about the importance, effects, and conceptualizations of well-being, and factors that may influence well-being. MethodThe issue comprises seven empirical studies, each exploring the relationship between teacher-related variables (e.g., emotions, growth mindset) or teaching-related variables (e.g., intercultural education, need-supportive teaching) and student well-being. It utilizes studies with methodologies ranging from large-scale assessments like the Program for International Student Assessment (PISA) to detailed longitudinal designs across various educational levels and contexts. ResultsFindings underscore the pivotal role of teaching in influencing student well-being. That is, specific teacher behaviors and pedagogical approaches were identified that correlate with student well-being. Key aspects included intercultural education, need-supportive teaching, and positive teacher–student relationships. The methodological strengths of the included studies were their use of large datasets, longitudinal designs, and multi-informant data. Based on these findings, I presented some suggestions for possible methodological improvements. ConclusionThis SI contained impressive articles that can be used as excellent models for research in this area. The findings can also expand our current knowledge by providing a multifaceted, multinational, and multi-informant understanding of the topic. Practical applications are suggested how teachers can support their students’ well-being and highlighted the importance of integrating well-being into educational discourse to provide a foundation for future investigations of teaching and learning processes.

A framework for in-field and out-of-field patient specific secondary cancer risk estimates from treatment plans using the TOPAS Monte Carlo system

Objective. To allow the estimation of secondary cancer risks from radiation therapy treatment plans in a comprehensive and user-friendly Monte Carlo (MC) framework. Method. Patient planning computed tomography scans were extended superior-inferior using the International Commission on Radiological Protection’s Publication 145 computational mesh phantoms and skeletal matching. Dose distributions were calculated with the TOPAS MC system using novel mesh capabilities and the digital imaging and communications in medicine radiotherapy extension interface. Finally, in-field and out-of-field cancer risk was calculated using both sarcoma and carcinoma risk models with two alternative parameter sets. Result. The TOPAS MC framework was extended to facilitate epidemiological studies on radiation-induced cancer risk. The framework is efficient and allows automated analysis of large datasets. Out-of-field organ dose was small compared to in-field dose, but the risk estimates indicate a non-negligible contribution to the total radiation induced cancer risk. Significance. This work equips the TOPAS MC system with anatomical extension, mesh geometry, and cancer risk model capabilities that make state-of-the-art out-of-field dose calculation and risk estimation accessible to a large pool of users. Furthermore, these capabilities will facilitate further refinement of risk models and sensitivity analysis of patient specific treatment options.

Molecular relaxation by reverse diffusion with time step prediction

Abstract Molecular relaxation, finding the equilibrium state of a non-equilibrium structure, is an essential component of computational chemistry to understand reactivity. Classical force field (FF) methods often rely on insufficient local energy minimization, while neural network FF models require large labeled datasets encompassing both equilibrium and non-equilibrium structures. As a remedy, we propose MoreRed, molecular relaxation by reverse diffusion, a conceptually novel and purely statistical approach where non-equilibrium structures are treated as noisy instances of their corresponding equilibrium states. To enable the denoising of arbitrarily noisy inputs via a generative diffusion model, we further introduce a novel diffusion time step predictor. Notably, MoreRed learns a simpler pseudo potential energy surface (PES) instead of the complex physical PES. It is trained on a significantly smaller, and thus computationally cheaper, dataset consisting of solely unlabeled equilibrium structures, avoiding the computation of non-equilibrium structures altogether. We compare MoreRed to classical FFs, equivariant neural network FFs trained on a large dataset of equilibrium and non-equilibrium data, as well as a semi-empirical tight-binding model. To assess this quantitatively, we evaluate the root-mean-square deviation between the found equilibrium structures and the reference equilibrium structures as well as their energies.

Image forgery detection by combining Visual Transformer with Variational Autoencoder Network

Recently, the applications and artificial intelligences used for image manipulation have become quite successful. In this case, the manipulation of personal data can lead to problems of insurmountable magnitude. Such problems not only put personal data at risk, but also lead us to unethical practices, with potentially irreversible negative consequences. For this reason, the reliability of image or video data is highly questionable. To solve this challenging problem, we introduce a Visual Transformer based Visual Transformer with Variational Autoencoder Network (ViT-VAE Net) model. The model includes Visual Transformer, one of the state-of-the-art architectures. In addition to this architecture, a Variational Auto Encoder structure is also included. is much more effective than models developed with the classical Convolutional Neural Network (CNN). Unlike models developed with CNN, it can perform operations on images of any size without being bound by a standard image resolution. In addition, thanks to the self-attention mechanism in the Visual Transformer architecture, manipulations on the image are caught more easily than CNN. The ViT-VAE Net model was trained with a large dataset and tested with 4 different datasets. With a success rate of 67 % on the training dataset, the model provided promising results. Very high rates were also obtained with the test datasets.

Inclusion of Shading and Soiling With Physical and Data-Driven Algorithms for Solar Power Forecasting

Shading and soiling are the biggest environmental factors that negatively affect the yield of PV systems. In order to integrate PV systems into the grid as easily and on large scale as possible, it is important that energy generation forecasts are as accurate as possible. The scope of this paper is to present a method how shading and soiling can be integrated into machine learning based PV forecasts even if they have already been pre trained by a large dataset. This paper focuses on shading by buildings, trees, obstacles, while shading by clouds can only be considered to a limited extent by weather forecasts. This study uses a dataset of three years of training data to build a base model. Subsequently, the power loss due to shading and soiling is determined using a digital twin and used to correct the forecast values of the baseline model. Finally, an evaluation of the corrected and original predicted values is performed. This study is able to show that the forecast error could be reduced in the same way as the loss due to shading and soiling using various machine learning methods. The results were compared against a Physical Informed Neural Network (PINN), which outperformed classical machine learning methods both with and without shading and soiling.

Proactive threat hunting to detect persistent behaviour-based advanced adversaries

Persistence behavior is a tactic advanced adversaries use to maintain unauthorized access and control of compromised assets over extended periods. Organizations can efficiently detect persistent adversaries and reduce the growing risks posed by highly skilled cyber threats by embracing creative techniques and utilizing sophisticated tools. By taking a proactive stance, businesses may increase their entire cybersecurity posture by anticipating and mitigating possible risks before they escalate. Security analysts perform thorough investigations and extract meaningful insights from large datasets with greater technical advantage by using Elasticsearch in conjunction with a variety of linguistic tools. This research presents a novel methodology for proactive threat intelligence to identify and mitigate advanced adversaries that use persistent behaviors. The authors designed and set up an Elasticsearch-based advanced Security Information and Event Management platform to offer a proactive threat-hunting strategy. This enables comprehensive analysis and detection by integrating Lucene, Kibana, and domain-specific languages. The goal of this research is to locate hidden advanced enemies who exhibit persistent behavior during cyberattacks. The framework can help improve the organization’s resilience to identify and respond to threats by closely examining activities like boot or logon auto-start execution in registry keys, tampering with system processes and services, and unauthorized creation of local accounts on compromised assets. This study emphasizes proactive actions over reactive reactions, which advances danger detection techniques. This technical study provides security practitioners seeking to improve defenses against new advanced attacks to stay ahead in a dynamic threat landscape.

Ant-colony optimization-based multi-input multi-output (ACO-MIMO) equalization for low-complexity mode-division multiplexing (MDM) processing

In mode-division multiplexing (MDM) systems, the computational complexity of the multi-input multi-output (MIMO) equalization module is a critical obstacle to practical development. The step size μ and the number of taps K are key parameters in the equalization algorithm, influencing the performance of finite impulse response (FIR) equalizers, including convergence speed and output signal quality. To alleviate the computational burden of locating the optimal μ-K combination, we propose two ant colony optimization (ACO) -based MIMO equalization schemes: the fixed ACO-MIMO and the random ACO-MIMO, corresponding to two optimization strategies. These schemes expedite the initialization process of both parameters. Subsequently, we conduct experiments to evaluate their performance in a 3-mode recirculating-loop transmission system. Our findings demonstrate that, compared to conventional schemes, such as genetic algorithm (GA) and steepest descent algorithm (SDA), the proposed ACO-MIMO schemes significantly reduce the number of calls to the equalization algorithm for locating optimal μ-K combination by up to 42.74% and 80.63%, reducing the complexity of the whole MIMO equalization for MDM systems. And the resulting hit-rate Phit for the optimal μ-K combination reaches up to 99.34%. Moreover, the ACO-MIMO schemes exhibit stable performance across different data collected from various round-trips, confirming the robust operation for the long-haul MDM transmission. Finally, we investigate the performance disparity between the two proposed ACO-MIMO schemes through bit-error-rate (BER) distribution, concluding that under a large dataset with various BER distributions, the performance of both schemes is essentially equivalent.

Hope and Its Distribution in Rural Tanzania

Recent research at the intersection of psychology and economics sheds light on the influence of hope on economic decisions. A body of that work concentrates on the economics of hope in developing country contexts. We identify two notable gaps: lack of attention to the measurement of hope as a latent psychological construct, and consequently, the lack of description and characterization of hope as a variable that can be measured and targeted. This study addresses these gaps by assessing the effectiveness of a novel hope measurement instrument, utilizing a large primary dataset collected in rural Tanzania. We estimate hope distributions across over 5,000 individuals and conditionally within subgroups defined by gender, region, recent shock, age, food security, income source, and religiosity. A positively-worded question about faith had the greatest information content among all questions, negatively worded questions were more effective in distinguishing people with relatively high hope. Employing generalized structural equation models, we observe significant variations in hope across sub-groups. Correcting for measurement distortions, we find significant heterogeneity in hope distributions across individuals and subgroups. The presence of an income-earning household member and religiosity yield the most pronounced shifts in hope distributions.

Predicting Behaviour Patterns in Online and PDF Magazines with AI Eye-Tracking

This study aims to improve college magazines, making them more engaging and user-friendly. We combined eye-tracking technology with artificial intelligence to accurately predict consumer behaviours and preferences. Our analysis included three college magazines, both online and in PDF format. We evaluated user experience using neuromarketing eye-tracking AI prediction software, trained on a large consumer neuroscience dataset of eye-tracking recordings from 180,000 participants, using Tobii X2 30 equipment, encompassing over 100 billion data points and 15 consumer contexts. An analysis was conducted with R programming v. 2023.06.0+421 and advanced SPSS statistics v. 27, IBM. (ANOVA, Welch’s Two-Sample t-test, and Pearson’s correlation). Our research demonstrated the potential of modern eye-tracking AI technologies in providing insights into various types of attention, including focus, engagement, cognitive demand, and clarity. The scientific accuracy of our findings, at 97–99%, underscores the reliability and robustness of our research, instilling confidence in the audience. This study also emphasizes the potential for future research to explore automated datasets, enhancing reliability and applicability across various fields and inspiring hope for further advancements in the field.

AI-Driven Threat Detection and Response Systems for Secure National Infrastructure Networks: A Comprehensive Review

Abstract: Due to the increased complexity and damage of cyberattacks in this digital age, the security of national infrastructure networks has become a vital concern. However, a possible approach to improve the cybersecurity of these crucial networks is to incorporate artificial intelligence (AI) into threat detection and response systems; to rapidly evaluate large data sets, identify anomalies, and automate countermeasures to lessen the effects of cyberattacks. The impact, implementation and approaches for anomaly detection and response automation of AI-powered solutions for safeguarding national infrastructure are examined in this paper. Understanding how AI technologies are used to automate threat detection and response, reviewing the operational usefulness of AI in enhancing cybersecurity measures and evaluating the deployment of these systems in critical infrastructure settings were also examined. The study revealed that the speed and accuracy of threat detection and response are greatly increased by AI-powered systems. The automation capacity of AI can potentially reduce the need for human analysts, while also providing faster threat mitigation. Additionally, the usefulness of AI across sectors indicates its practicality in situations and how it may adapt in response to new threats. In conclusion, AI-driven threat detection and response systems are an important development in national infrastructure network cybersecurity. Therefore, by improving the capacity to recognize and address cyber-attacks these technologies can ultimately increase the overall resilience of national infrastructures.

9 Delving into molecular underpinnings of sarcomatoid/rhabdoid dedifferentioation in renal cancers using spatial profiling

Abstract Background Renal cell carcinomas (RCC) are characterized by their largely diverse clinical outcomes. Leveraging genomic diversity among individual tumors, our research during the past decade has concentrated on developing personalized medicine approaches for the most common and the most aggressive type of RCC, clear cell renal carcinoma (ccRCC). We generated the largest dataset of ccRCC including somatic genomic and clinical annotations for over 940 ccRCC patients, and developed a genomic classifier, based on mutational status of 12 RCC-relevant genes, which is able to stratify patients according to their risk of relapse after nephrectomy and/or death due to RCC. Furthermore, we noticed the presence of mesenchymal-like cellular phenotypes in tumors of high-risk patients, representsing a de-differentiation process, known as sarcomatoic or rhabdoid (S/R) de-differentation. Methods The current knowledge about molecular mechanisms that may drive S/R de-differentiation is primarily generated from molecular profiling of bulk tumors, collecting data from a mixture of cells that co-exist in the tumor milieu. Therefore, high-resolution studies that precisely define molecular characteristics of different cellular phenotypes associated with S/R features are missing. We have used spatial transcriptomic profiling to investigate molecular mechanisms that underline S/R de-differentationin RCC. While preserving tissue context, we applied spatial whole human transcriptome profiling to areas exhibiting S/R or clear cell phenotypes within the same tumor specimens to generate phenotype-specific transcriptome profiles. In addition, we applied whole-exome sequencing (WES) to independent areas that exhibit different differentiation phenotype withing same tumors. Results Pathway and network analysis of genesets with upregulation in each area have revealed meaningful differences in cellular pathways which are active in each phenotype. Specifically, we observed that dysregulation of extracellular matrix (ECM) is a hallmark of S/R areas within RCC tumors. Furthermore, WES of phenotypically-distinct areas within each tumor has shed new light son the genome evoltuion of S/R phenotypes. Conclusions S/R dedifferentiation in RCC tumors is characterized by specific genomic evolutionary patterns and substantial dysregulation of ECM components. DOD CDMRP Funding: yes