Biological Networks Research Articles

Mass spectrum oriented metabolomics for evaluating the efficacy and discovering the mechanism of Shaofuzhuyu Decoction for endometriosis of cold coagulation and blood stasis

Shaofuzhuyu Decoction (SFZYD) is a classical formula for treating endometriosis of cold coagulation and blood stasis (ECB). The clinical efficacy is definite, but the potential mechanisms require further exploration. The study aimed to reveal the metabolic mechanisms of SFZYD for treating ECB using mass spectrum oriented metabolomics. Firstly, the study has used metabolomics data to identify biomarkers and to investigate metabolic pathways. Then, the targets of SFZYD for treating ECB were dug by building and analyzing a biological network of biomarkers. Finally, the obtained targets were validated by molecular docking. This study found that SFZYD could significantly improve the biochemical indicators and metabolic abnormalities of ECB. A total of 18 ECB-related biomarkers in 7 pathways were identified. SFZYD was able to regulate the levels of 14 biomarkers that were involved in 5 metabolic pathways. Furthermore, the study yielded 119 SFZYD active ingredients, 1119 target proteins associated with endometriosis, 610 targets associated with biomarkers, 727 GO functions, and 159 KEGG pathways. Biological network analysis constructed a network diagram of herbs-ingredients-targets-biomarkers, and found 6 key active ingredients and 9 core targets. Molecular docking showed high affinities between key ingredients and core targets. This study elucidated that SFZYD plays a role in treating ECB through multi-component, multi-target, and multi-pathway.

The difference in gene expression network based on gender in the recovery of knee cruciate ligament injuries during exercise

Anterior cruciate ligament (ACL) injuries are common and often debilitating, requiring extensive rehabilitation. The recovery process involves complex physiological, biomechanical, and biochemical factors. Gender differences in the incidence and recovery from ACL injuries are well-documented, with recent advances in molecular biology suggesting that gene expression networks play a crucial role in tissue repair and regeneration. This study aims to compare gene expression networks between genders to provide strategies for better-managing sports injuries. Gene expression data were sourced from the Gene Expression Omnibus (GEO) database, with samples divided into four groups based on gender and age. Differentially expressed genes (DEGs) were identified using GEO2R, and biological interaction networks were constructed using the STRING (Search Tool for the Retrieval of Interacting Genes/Proteins). Network analysis was performed using Gephi, focusing on betweenness centrality and overall centrality. Two common genes, TRNA Aspartic Acid Methyltransferase 1 (TRDMT1) and NCF1, were identified in both sexes but exhibited different centrality measures. NCF1, associated with the production of superoxide anion and linked to chronic granulomatous disease, was expressed ten times higher in men than women. TRDMT1, responsible for RNA methylation, was twice as highly expressed in men. The study underscores the importance of gender-specific molecular mechanisms in ACL injury recovery and highlights the need for personalized treatment strategies.

Identification of Critical Genes Differentiating Stable and Unstable Atherosclerotic Plaques: A Bioinformatic and Computational Analysis.

Identification of biomarkers to distinguish between stable and unstable plaque formation would be very useful to predict plaque vulnerability. We downloaded microarray profiles of gene set enrichment (GSE) accession numbers including GSE71226 and GSE20680 (group A: containing healthy vs stable plaque samples) and GSE62646 and GSE34822 (group B: containing stable vs unstable plaque samples) from Gene expression omnibus (GEO) database. Differentially expressed genes were compared in both data sets of each group. Ten and 12 key genes were screened in groups A and B, respectively. Gene Ontology (GO) enrichment was applied by the plugin "BiNGO" (Biological networks gene ontology tool) of the Cytoscape. The key genes were mostly enriched in the biological process of positive regulation of the cellular process. The protein-protein interaction and co-expression network were analyzed by the STRING (search tool for the retrieval of interacting genes/proteins) and GeneMANIA (gene multiple association network integration algorithm) plugin of Cytoscape, respectively, which showed that Epidermal growth factor (EGF), Heparin-binding EGF like growth factor (HBEGF), and Matrix metalloproteinase 9 (MMP9) were at the core of the network. Further validation of key genes using two datasets showed that Phosphodiesterase 5A (PDE5A) and Protein S (PROS1) were decreased in unstable plaques, while Suppressor of cytokine signaling (SOCS3), HBEGF, and Leukocyte immunoglobulin-like receptor B4 (LILRB4) were increased. The present study used several datasets to identify key genes associated with stable and unstable atherosclerotic plaque.

Biological regulatory network analysis for targeting the mitochondrial calcium uniporter (MCU) mediated calcium (Ca2+) transport in neurodegenerative disorders

AbstractCalcium (Ca2+) has been observed as the most important ion involved in a series of cellular processes and its homeostasis is critical for normal cellular functions. Mitochondrial calcium uniporter (MCU) complex has been recognized as the most important calcium‐specific channel located in the inner mitochondrial membrane and is one of the major players in maintaining the Ca2+ homeostasis by transporting Ca2+ across the mitochondrial membrane. Furthermore, dysregulation of the mitochondrial Ca2+ homeostasis has been orchestrated to neurodegenerative response. This necessitates quantitative evaluation of the MCU‐dependent mROS production and subsequent cellular responses for more specific therapeutic interventions against neurodegenerative disorders. Towards this goal, here we present a biological regulatory network of MCU to dynamically simulate the MCU‐mediated ROS production and its response in neurodegeneration. Previously, ruthenium complex RuRed and its derivatives have been reported to show low nM to high µM potency against MCU to maintain cytosolic Ca2+ (cCa2+) homeostasis by modulating mitochondrial Ca2+ (mCa2+) uptake. Therefore, structural modeling and dynamic simulation of MCU pore‐forming subunit is performed to probe the interaction profiling of previously reported Ru265 and its derivatives compounds with MCU. The current study highlighted MCU as a potential drug target in neurodegenerative disorders. Furthermore, ASP261 and GLU264 amino acid residues in DIME motif of MCU pore‐forming subunits are identified as crucial for modulating the activity of MCU in neurodegenerative disorders.

Immunological Signatures for Early Detection of Human Head and Neck Squamous Cell Carcinoma through RNA Transcriptome Analysis of Blood Platelets.

Although there has been a reduction in head and neck squamous cell carcinoma occurrence, it continues to be a serious global health concern. The lack of precise early diagnostic biomarkers and postponed diagnosis in the later stages are notable constraints that contribute to poor survival rates and emphasize the need for innovative diagnostic methods. In this study, we employed machine learning alongside weighted gene co-expression network analysis (WGCNA) and network biology to investigate the gene expression patterns of blood platelets, identifying transcriptomic markers for HNSCC diagnosis. Our comprehensive examination of publicly available gene expression datasets revealed nine genes with significantly elevated expression in samples from individuals diagnosed with HNSCC. These potential diagnostic markers were further assessed using TCGA and GTEx datasets, demonstrating high accuracy in distinguishing between HNSCC and non-cancerous samples. The findings indicate that these gene signatures could revolutionize early HNSCC identification. Additionally, the study highlights the significance of tumor-educated platelets (TEPs), which carry RNA signatures indicative of tumor-derived material, offering a non-invasive source for early-detection biomarkers. Despite using platelet and tumor samples from different individuals, our results suggest that TEPs reflect the transcriptomic and epigenetic landscape of tumors. Future research should aim to directly correlate tumor and platelet samples from the same patients to further elucidate this relationship. This study underscores the potential of these biomarkers in transforming early diagnosis and personalized treatment strategies for HNSCC, advocating for further research to validate their predictive and therapeutic potential.

PROSPECTS AND LIMITS OF ARTIFICIAL INTELLIGENCE IN THE CONTEXT OF NATURAL INTELLIGENCE (philosophical and methodological aspects) / ПЕРСПЕКТИВЫ И ПРЕДЕЛЫ ВОЗМОЖНОСТЕЙ ИСКУССТВЕННОГО ИНТЕЛЛЕКТА В КОНТЕКСТЕ ЕСТЕСТВЕННОГО ИНТЕЛЛЕКТА (философско-методологические аспекты)

Artificial intelligence is characterized as non-living intelligence, although attempts are being made to pass from the principles of biological neural networks to artificial living neural networks. The human brain is assumed to be a complex system of physicochemical neurophysiological biochemical integration of neural networks, which encodes a specific type of biological information, cognitive information. Intelligence is a specific quality of the process of thinking as an expression based on the neurophysiological mechanism of prediction. Nowadays, people more and more often speak about the anthropological, axiological crisis, modification and transformation of human nature, the creation of biological hybrids that compete with natural ones. It will hardly be possible to formalize the complexities of the human brain and mind into digital code in artificial intelligence.

Transfer learning of condition-specific perturbation in gene interactions improves drug response prediction.

Drug response is conventionally measured at the cell level, often quantified by metrics like IC50. However, to gain a deeper understanding of drug response, cellular outcomes need to be understood in terms of pathway perturbation. This perspective leads us to recognize a challenge posed by the gap between two widely used large-scale databases, LINCS L1000 and GDSC, measuring drug response at different levels-L1000 captures information at the gene expression level, while GDSC operates at the cell line level. Our study aims to bridge this gap by integrating the two databases through transfer learning, focusing on condition-specific perturbations in gene interactions from L1000 to interpret drug response integrating both gene and cell levels in GDSC. This transfer learning strategy involves pretraining on the transcriptomic-level L1000 dataset, with parameter-frozen fine-tuning to cell line-level drug response. Our novel condition-specific gene-gene attention (CSG2A) mechanism dynamically learns gene interactions specific to input conditions, guided by both data and biological network priors. The CSG2A network, equipped with transfer learning strategy, achieves state-of-the-art performance in cell line-level drug response prediction. In two case studies, well-known mechanisms of drugs are well represented in both the learned gene-gene attention and the predicted transcriptomic profiles. This alignment supports the modeling power in terms of interpretability and biological relevance. Furthermore, our model's unique capacity to capture drug response in terms of both pathway perturbation and cell viability extends predictions to the patient level using TCGA data, demonstrating its expressive power obtained from both gene and cell levels. The source code for the CSG2A network is available at https://github.com/eugenebang/CSG2A.

Neurodevelopmental Disorders Associated with Gut Microbiome Dysbiosis in Children.

The formation of the human gut microbiome initiates in utero, and its maturation is established during the first 2-3 years of life. Numerous factors alter the composition of the gut microbiome and its functions, including mode of delivery, early onset of breastfeeding, exposure to antibiotics and chemicals, and maternal stress, among others. The gut microbiome-brain axis refers to the interconnection of biological networks that allow bidirectional communication between the gut microbiome and the brain, involving the nervous, endocrine, and immune systems. Evidence suggests that the gut microbiome and its metabolic byproducts are actively implicated in the regulation of the early brain development. Any disturbance during this stage may adversely affect brain functions, resulting in a variety of neurodevelopmental disorders (NDDs). In the present study, we reviewed recent evidence regarding the impact of the gut microbiome on early brain development, alongside its correlation with significant NDDs, such as autism spectrum disorder, attention-deficit/hyperactivity disorder, Tourette syndrome, cerebral palsy, fetal alcohol spectrum disorders, and genetic NDDs (Rett, Down, Angelman, and Turner syndromes). Understanding changes in the gut microbiome in NDDs may provide new chances for their treatment in the future.

Prioritizing drug targets by perturbing biological network response functions.

Therapeutic interventions are designed to perturb the function of a biological system. However, there are many types of proteins that cannot be targeted with conventional small molecule drugs. Accordingly, many identified gene-regulatory drivers and downstream effectors are currently undruggable. Drivers and effectors are often connected by druggable signaling and regulatory intermediates. Methods to identify druggable intermediates therefore have general value in expanding the set of targets available for hypothesis-driven validation. Here we identify and prioritize potential druggable intermediates by developing a network perturbation theory, termed NetPert, for response functions of biological networks. Dynamics are defined by a network structure in which vertices represent genes and proteins, and edges represent gene-regulatory interactions and protein-protein interactions. Perturbation theory for network dynamics prioritizes targets that interfere with signaling from driver to response genes. Applications to organoid models for metastatic breast cancer demonstrate the ability of this mathematical framework to identify and prioritize druggable intermediates. While the short-time limit of the perturbation theory resembles betweenness centrality, NetPert is superior in generating target rankings that correlate with previous wet-lab assays and are more robust to incomplete or noisy network data. NetPert also performs better than a related graph diffusion approach. Wet-lab assays demonstrate that drugs for targets identified by NetPert, including targets that are not themselves differentially expressed, are active in suppressing additional metastatic phenotypes.

The centrosomal protein FGFR1OP controls myosin function in murine intestinal epithelial cells

Recent advances in human genetics have shed light on the genetic factors contributing to inflammatory diseases, particularly Crohn's disease (CD), a prominent form of inflammatory bowel disease. Certain risk genes associated with CD directly influence cytokine biology and cell-specific communication networks. Current CD therapies primarily rely on anti-inflammatory drugs, which are inconsistently effective and lack strategies for promoting epithelial restoration and mucosal balance. To understand CD's underlying mechanisms, weinvestigated the link between CD and the FGFR1OP gene, which encodes a centrosome protein. FGFR1OP deletion in mouse intestinal epithelial cells disrupted crypt architecture, resulting in crypt loss, inflammation, and fatality. FGFR1OP insufficiency hindered epithelial resilience during colitis. FGFR1OP was crucial for preserving non-muscle myosin II activity, ensuring the integrity of the actomyosin cytoskeleton and crypt cell adhesion. This role of FGFR1OP suggests that its deficiency in genetically predisposed individuals may reduce epithelial renewal capacity, heightening susceptibility to inflammation and disease.

Multistability and synchronicity of memristor coupled adaptive synaptic neuronal network

Memristor coupled neuronal networks can simulate the complex interactions between neurons in biological neuronal networks. These capture the current and synaptic transmission between neurons through mathematical equations, thus reflecting the signal transmission and integration in biological neuronal networks. This paper presents a five-dimensional isomorphic memristor coupled adaptive synaptic neuron (mCASN) network using a non-ideal memristor to couple two-dimensional adaptive synaptic neurons. The theoretical analysis of equilibrium points shows that the mCASN network has many different stability types. The complex parameters-dependent bifurcation behaviors are studied using various numerical methods, and the initial state-dependent coexistence behaviors and the riddled-like basins of attraction sensitive to initial states are further revealed, which proves that the coupled network has multistability. In addition, using the normalized average synchronization error method, we reveal that the mCASN network has synchronization characteristics controlled by parameters and initial states. The complete synchronization and lag synchronization induced by the network parameters, the internal variable of the coupled memristor, and the state variable of the sub-neuron are numerically analyzed. Finally, the mCASN network is implemented using an MCU-based digital hardware platform, and the numerical results are verified.

A Network Toxicology Approach for Mechanistic Modelling of Nanomaterial Hazard and Adverse Outcomes.

Hazard assessment is the first step in evaluating the potential adverse effects of chemicals. Traditionally, toxicological assessment has focused on the exposure, overlooking the impact of the exposed system on the observed toxicity. However, systems toxicology emphasizes how system properties significantly contribute to the observed response. Hence, systems theory states that interactions store more information than individual elements, leading to the adoption of network based models to represent complex systems in many fields of life sciences. Here, they develop a network-based approach to characterize toxicological responses in the context of a biological system, inferring biological system specific networks. They directly link molecular alterations to the adverse outcome pathway (AOP) framework, establishing direct connections between omics data and toxicologically relevant phenotypic events. They apply this framework to a dataset including 31 engineered nanomaterials with different physicochemical properties in two different in vitro and one in vivo models and demonstrate how the biological system is the driving force of the observed response. This work highlights the potential of network-based methods to significantly improve their understanding of toxicological mechanisms from a systems biology perspective and provides relevant considerations and future data-driven approaches for the hazard assessment of nanomaterials and other advanced materials.

Representations and generalization in artificial and brain neural networks

Humans and animals excel at generalizing from limited data, a capability yet to be fully replicated in artificial intelligence. This perspective investigates generalization in biological and artificial deep neural networks (DNNs), in both in-distribution and out-of-distribution contexts. We introduce two hypotheses: First, the geometric properties of the neural manifolds associated with discrete cognitive entities, such as objects, words, and concepts, are powerful order parameters. They link the neural substrate to the generalization capabilities and provide a unified methodology bridging gaps between neuroscience, machine learning, and cognitive science. We overview recent progress in studying the geometry of neural manifolds, particularly in visual object recognition, and discuss theories connecting manifold dimension and radius to generalization capacity. Second, we suggest that the theory of learning in wide DNNs, especially in the thermodynamic limit, provides mechanistic insights into the learning processes generating desired neural representational geometries and generalization. This includes the role of weight norm regularization, network architecture, and hyper-parameters. We will explore recent advances in this theory and ongoing challenges. We also discuss the dynamics of learning and its relevance to the issue of representational drift in the brain.

Mechanistic insights into xanthomicrol as the active anti-HCC ingredient of Phytolacca acinosa Roxb.: A network pharmacology analysis and transcriptomics integrated experimental verification

Ethnopharmacological relevancePhytolacca acinosa Roxb. (PAR) is a Traditional Chinese Medicinal (TCM) plant with a broad global distribution encompassing 35 species, four of which are found in the People's Republic of China. It occupies a significant role in both Oriental and American traditional medicine, employed in treating a range of conditions such as edema, inflammation, dermatitis, and rheumatism. PAR is also used as a molluscicide and for addressing tumors and bronchitis. The plant is documented in the Chinese Pharmacopoeia and has a longstanding history in TCM, particularly for its diuretic properties and in treating ailments such as edema, swelling, and ulcers. Notably, PAR has demonstrated potent inhibitory effects against the A549 human lung cancer cell line, underscoring its potential in contributing to the development of novel cancer therapeutics. Aim of the studyThe research aims to elucidate the active components of PAR and their mechanisms in treating hepatocellular carcinoma (HCC). Materials and methodsEmploying network pharmacology, this study predicted the principal active compounds and key targets of PAR. A holistic methodology incorporating biological network analysis, transcriptomics sequencing, molecular docking, and molecular dynamics (MD) simulations was utilized to forecast the effects of PAR on HCC, with empirical evidence supporting these findings. ResultsNetwork pharmacology identified xanthomicrol as the foremost active compound in PAR. The tumor-suppressive functions of PAR, as indicated by KEGG pathway analysis and transcriptomics sequencing, predominantly occur via the PI3K/AKT pathway. Molecular docking and dynamics simulations demonstrated the high affinity of xanthomicrol towards TNF, MMP9, PPARG, KDR, and MMP2. In vivo experiments verified the efficacy of xanthomicrol in curtailing HCC tumor growth, while in vitro assessments revealed its substantial impact on the proliferation and apoptosis of HepG2 and HCCLM3 cells. Moreover, the study indicates that xanthomicrol may modulate the expression of TNF, MMP9, PPARG, KDR, and MMP2 in HCC cells and inhibit the activation of the PI3K/AKT pathway. ConclusionsXanthomicrol, a principal active component of PAR, has been identified to impede the growth of HCC by targeting the PI3K/Akt/MMP9 pathway. This insight could enhance therapeutic approaches for HCC.

An Artificial Universal Tactile Nociceptor Based on 2D Polymer Film Memristor Arrays with Tunable Resistance Switching Behaviors.

Nociceptor is an important receptor in the organism's sensory system; it can perceive harmful stimuli and send signals to the brain in order to protect the body in time. The injury degree of nociceptor can be divided into three stages: self-healing injury, treatable injury, and permanent injury. However, the current studies on nociceptor simulation are limited to the self-healing stage due to the limitation of the untunable resistance switching behavior of memristors. In this study, we constructed Al/2DPTPAK+TAPB/Ag memristor arrays with adjustable memory behaviors to emulate the nociceptor of biological neural network of all three stages. For this purpose, a PDMS/AgNWs/ITO/PET pressure sensor was assembled to mimic the tactile perception of the skin. The memristor arrays can not only simulate all the response of nociceptor, i.e., the threshold, relaxation, no adaptation, and sensitization with the self-healing injury, but can also simulate the treatable injury and the permanent injury. These behaviors are both demonstrated with a single memristor and in the form of pattern mapping of the memristor array.

Biological direct-shortcut deep residual learning for sparse visual processing

We show, based on the following three grounds, that the primary visual cortex (V1) is a biological direct-shortcut deep residual learning neural network (ResNet) for sparse visual processing: (1) We first highlight that Gabor-like sets of basis functions, which are similar to the receptive fields of simple cells in the primary visual cortex (V1), are excellent candidates for sparse representation of natural images; i.e., images from the natural world, affirming the brain to be optimized for this. (2) We then prove that the intra-layer synaptic weight matrices of this region can be reasonably first-order approximated by identity mappings, and are thus sparse themselves. (3) Finally, we point out that intra-layer weight matrices having identity mapping as their initial approximation, irrespective of this approximation being also a reasonable first-order one or not, resemble the building blocks of direct-shortcut digital ResNets, which completes the grounds. This biological ResNet interconnects the sparsity of the final representation of the image to that of its intra-layer weights. Further exploration of this ResNet, and understanding the joint effects of its architecture and learning rules, e.g. on its inductive bias, could lead to major advancements in the area of bio-inspired digital ResNets. One immediate line of research in this context, for instance, is to study the impact of forcing the direct-shortcuts to be good first-order approximations of each building block. For this, along with the ℓ1\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$${{\\ell}}_{1}$$\\end{document}-minimization posed on the basis function coefficients the ResNet finally provides at its output, another parallel one could e.g. also be posed on the weights of its residual layers.

Biological network comparison identifies a novel synergistic mechanism of Ginseng Radix-Astragali Radix herb pair in cancer-related fatigue

Ethnopharmacological relevanceGinseng Radix and Astragali Radix are commonly combined to tonify Qi and alleviate fatigue. Previous studies have employed biological networks to investigate the mechanisms of herb pairs in treating different diseases. However, these studies have only elucidated a single network for each herb pair, without emphasizing the superiority of the herb combination over individual herbs. Aim of the studyThis study proposes an approach of comparing biological networks to highlight the synergistic effect of the pair in treating cancer-related fatigue (CRF). MethodsThe compounds and targets of Ginseng Radix, Astragali Radix, and CRF diseases were collected and predicted using different databases. Subsequently, the overlapping targets between herbs and disease were imported into the STRING and DAVID tools to build protein-protein interaction (PPI) networks and analyze enriched KEGG pathways. The biological networks of Ginseng Radix and Astragali Radix were compared separately or together using the DyNet application. Molecular docking was used to verify the predicted results. Further, in vitro experiments were conducted to validate the synergistic pathways identified in in silico studies. ResultsIn the PPI network comparison, the combination created 89 new interactions and an increased average degree (11.260) when compared to single herbs (10.296 and 9.394). The new interactions concentrated on HRAS, STAT3, JUN, and IL6. The topological analysis identified 20 core targets of the combination, including three Ginseng Radix-specific targets, three Astragali Radix-specific targets, and 14 shared targets. In KEGG enrichment analysis, the combination regulated additional signaling pathways (152) more than Ginseng Radix (146) and Astragali Radix (134) alone. The targets of the herb pair synergistically regulated cancer pathways, specifically hypoxia-inducible factor 1 (HIF-1) signaling pathway. In vitro experiments including enzyme-linked immunosorbent assay and Western blot demonstrated that two herbs combination could up-regulate HIF-1α signaling pathway at different combined concentrations compared to either single herb alone. ConclusionThe herb pair increased protein interactions and adjusted metabolic pathways more than single herbs. This study provides insights into the combination of Ginseng Radix and Astragali Radix in clinical practice.

Target repositioning using multi-layer networks and machine learning: The case of prostate cancer

The discovery of novel therapeutic targets, defined as proteins which drugs can interact with to induce therapeutic benefits, typically represent the first and most important step of drug discovery. One solution for target discovery is target repositioning, a strategy which relies on the repurposing of known targets for new diseases, leading to new treatments, less side effects and potential drug synergies. Biological networks have emerged as powerful tools for integrating heterogeneous data and facilitating the prediction of biological or therapeutic properties. Consequently, they are widely employed to predict new therapeutic targets by characterizing potential candidates, often based on their interactions within a Protein-Protein Interaction (PPI) network, and their proximity to genes associated with the disease. However, over-reliance on PPI networks and the assumption that potential targets are necessarily near known genes can introduce biases that may limit the effectiveness of these methods. This study addresses these limitations in two ways. First, by exploiting a multi-layer network which incorporates additional information such as gene regulation, metabolite interactions, metabolic pathways, and several disease signatures such as Differentially Expressed Genes, mutated genes, Copy Number Alteration, and structural variants. Second, by extracting relevant features from the network using several approaches including proximity to disease-associated genes, but also unbiased approaches such as propagation-based methods, topological metrics, and module detection algorithms. Using prostate cancer as a case study, the best features were identified and utilized to train machine learning algorithms to predict 5 novel promising therapeutic targets for prostate cancer: IGF2R, C5AR, RAB7, SETD2 and NPBWR1.

Discovering neural policies to drive behaviour by integrating deep reinforcement learning agents with biological neural networks

Discovering neural policies to drive behaviour by integrating deep reinforcement learning agents with biological neural networks

Exploiting Signal Propagation Delays to Match Task Memory Requirements in Reservoir Computing.

Recurrent neural networks (RNNs) transmit information over time through recurrent connections. In contrast, biological neural networks use many other temporal processing mechanisms. One of these mechanisms is the inter-neuron delays caused by varying axon properties. Recently, this feature was implemented in echo state networks (ESNs), a type of RNN, by assigning spatial locations to neurons and introducing distance-dependent inter-neuron delays. These delays were shown to significantly improve ESN task performance. However, thus far, it is still unclear why distance-based delay networks (DDNs) perform better than ESNs. In this paper, we show that by optimizing inter-node delays, the memory capacity of the network matches the memory requirements of the task. As such, networks concentrate their memory capabilities to the points in the past which contain the most information for the task at hand. Moreover, we show that DDNs have a greater total linear memory capacity, with the same amount of non-linear processing power.