Ana B. Porto-Pazos

Artificial astrocytes improve neural network performance

Compelling evidence indicates the existence of bidirectional communication between astrocytes and neurons. Astrocytes, a type of glial cells classically considered to be passive supportive cells, have been recently demonstrated to be actively involved in the processing and regulation of synaptic information, suggesting that brain function arises from the activity of neuron-glia networks. However, the actual impact of astrocytes in neural network function is largely unknown and its application in artificial intelligence remains untested. We have investigated the consequences of including artificial astrocytes, which present the biologically defined properties involved in astrocyte-neuron communication, on artificial neural network performance. Using connectionist systems and evolutionary algorithms, we have compared the performance of artificial neural networks (NN) and artificial neuron-glia networks (NGN) to solve classification problems. We show that the degree of success of NGN is superior to NN. Analysis of performances of NN with different number of neurons or different architectures indicate that the effects of NGN cannot be accounted for an increased number of network elements, but rather they are specifically due to astrocytes. Furthermore, the relative efficacy of NGN vs. NN increases as the complexity of the network increases. These results indicate that artificial astrocytes improve neural network performance, and established the concept of Artificial Neuron-Glia Networks, which represents a novel concept in Artificial Intelligence with implications in computational science as well as in the understanding of brain function.

Artificial neuron–glia networks learning approach based on cooperative coevolution

Artificial Neuron-Glia Networks (ANGNs) are a novel bio-inspired machine learning approach. They extend classical Artificial Neural Networks (ANNs) by incorporating recent findings and suppositions about the way information is processed by neural and astrocytic networks in the most evolved living organisms. Although ANGNs are not a consolidated method, their performance against the traditional approach, i.e. without artificial astrocytes, was already demonstrated on classification problems. However, the corresponding learning algorithms developed so far strongly depends on a set of glial parameters which are manually tuned for each specific problem. As a consequence, previous experimental tests have to be done in order to determine an adequate set of values, making such manual parameter configuration time-consuming, error-prone, biased and problem dependent. Thus, in this paper, we propose a novel learning approach for ANGNs that fully automates the learning process, and gives the possibility of testing any kind of reasonable parameter configuration for each specific problem. This new learning algorithm, based on coevolutionary genetic algorithms, is able to properly learn all the ANGNs parameters. Its performance is tested on five classification problems achieving significantly better results than ANGN and competitive results with ANN approaches.

Artificial Intelligence Techniques for Colorectal Cancer Drug Metabolism: Ontologies and Complex Networks

Colorectal cancer is one of the most frequent types of cancer in the world and generates important social impact. The understanding of the specific metabolism of this disease and the transformations of the specific drugs will allow finding effective prevention, diagnosis and treatment of the colorectal cancer. All the terms that describe the drug metabolism contribute to the construction of ontology in order to help scientists to link the correlated information and to find the most useful data about this topic. The molecular components involved in this metabolism are included in complex network such as metabolic pathways in order to describe all the molecular interactions in the colorectal cancer. The graphical method of processing biological information such as graphs and complex networks leads to the numerical characterization of the colorectal cancer drug metabolic network by using invariant values named topological indices. Thus, this method can help scientists to study the most important elements in the metabolic pathways and the dynamics of the networks during mutations, denaturation or evolution for any type of disease. This review presents the last studies regarding ontology and complex networks of the colorectal cancer drug metabolism and a basic topology characterization of the drug metabolic process sub-ontology from the Gene Ontology.

Computational models of neuron-astrocyte interactions lead to improved efficacy in the performance of neural networks.

The importance of astrocytes, one part of the glial system, for information processing in the brain has recently been demonstrated. Regarding information processing in multilayer connectionist systems, it has been shown that systems which include artificial neurons and astrocytes (Artificial Neuron-Glia Networks) have well-known advantages over identical systems including only artificial neurons. Since the actual impact of astrocytes in neural network function is unknown, we have investigated, using computational models, different astrocyte-neuron interactions for information processing; different neuron-glia algorithms have been implemented for training and validation of multilayer Artificial Neuron-Glia Networks oriented toward classification problem resolution. The results of the tests performed suggest that all the algorithms modelling astrocyte-induced synaptic potentiation improved artificial neural network performance, but their efficacy depended on the complexity of the problem.

Deep artificial neural networks and neuromorphic chips for big data analysis: pharmaceutical and bioinformatics applications

Over the past decade, Deep Artificial Neural Networks (DNNs) have become the state-of-the-art algorithms in Machine Learning (ML), speech recognition, computer vision, natural language processing and many other tasks. This was made possible by the advancement in Big Data, Deep Learning (DL) and drastically increased chip processing abilities, especially general-purpose graphical processing units (GPGPUs). All this has created a growing interest in making the most of the potential offered by DNNs in almost every field. An overview of the main architectures of DNNs, and their usefulness in Pharmacology and Bioinformatics are presented in this work. The featured applications are: drug design, virtual screening (VS), Quantitative Structure–Activity Relationship (QSAR) research, protein structure prediction and genomics (and other omics) data mining. The future need of neuromorphic hardware for DNNs is also discussed, and the two most advanced chips are reviewed: IBM TrueNorth and SpiNNaker. In addition, this review points out the importance of considering not only neurons, as DNNs and neuromorphic chips should also include glial cells, given the proven importance of astrocytes, a type of glial cell which contributes to information processing in the brain. The Deep Artificial Neuron–Astrocyte Networks (DANAN) could overcome the difficulties in architecture design, learning process and scalability of the current ML methods.

Ontologies of drug discovery and design for neurology, cardiology and oncology.

The complex diseases in the field of Neurology, Cardiology and Oncology have the most important impact on our society. The theoretical methods are fast and they involve some efficient tools aimed at discovering new active drugs specially designed for these diseases. The ontology of all the items that are linked with the molecule metabolism and the treatment of these diseases gives us the possibility to correlate information from different levels and to discover new relationships between complex diseases such as common drug targets and disease patterns. This review presents the ontologies used to process drug discovery and design in the most common complex diseases.

Parallel Computing for Brain Simulation

BACKGROUND The human brain is the most complex system in the known universe, it is therefore one of the greatest mysteries. It provides human beings with extraordinary abilities. However, until now it has not been understood yet how and why most of these abilities are produced. AIMS For decades, researchers have been trying to make computers reproduce these abilities, focusing on both understanding the nervous system and, on processing data in a more efficient way than before. Their aim is to make computers process information similarly to the brain. Important technological developments and vast multidisciplinary projects have allowed creating the first simulation with a number of neurons similar to that of a human brain. CONCLUSION This paper presents an up-to-date review about the main research projects that are trying to simulate and/or emulate the human brain. They employ different types of computational models using parallel computing: digital models, analog models and hybrid models. This review includes the current applications of these works, as well as future trends. It is focused on various works that look for advanced progress in Neuroscience and still others which seek new discoveries in Computer Science (neuromorphic hardware, machine learning techniques). Their most outstanding characteristics are summarized and the latest advances and future plans are presented. In addition, this review points out the importance of considering not only neurons: Computational models of the brain should also include glial cells, given the proven importance of astrocytes in information processing.

The Ability of MEAs Containing Cultured Neuroglial Networks to Process Information

The study of the nervous system of human beings is an arduous task. The reasons are that it is very complex and it is internal to the organism. The nervous system is comprised not only of neuronal networks but also of different types of cells that constitute the glial system. Astrocytes, a type of glial cells, have traditionally been considered as passive, supportive cells. However, through the use of neuroscientific techniques, it has recently been demonstrated that astrocytes are actively involved in the processing and regulation of synaptic information, suggesting that brain function arises from the activity of neuron-glia networks. Also in recent studies employing artificial intelligence (AI) techniques, it has been shown that adding artificial astrocytes to Artificial Neural Networks (ANNs), the effectiveness of such networks in classification tasks is markedly improved. At present, the actual impact of astrocytes in neural network function is largely unknown. Therefore, our group is placing increasing emphasis on the study of the influence that astrocytes may have on brain information processing using a rather different perspective based on the use of multielectrode arrays (MEAs). This represents a hybrid approach given that it combines a biological component (cultured cells), hardware technology (MEAs), and AI (computer simulations based on AI techniques to control the system). With this in mind, the objective of this paper is to present a review of the state of the art in the use of MEAs containing nerve cells. This review is intended as a preliminary theoretical analysis on the suitability of these devices to achieve the aforementioned future goal of fusing bioinformatics, micro/nano-technologies, and AI techniques to study these complex systems.

Prediction of high anti-angiogenic activity peptides in silico using a generalized linear model and feature selection

Screening and in silico modeling are critical activities for the reduction of experimental costs. They also speed up research notably and strengthen the theoretical framework, thus allowing researchers to numerically quantify the importance of a particular subset of information. For example, in fields such as cancer and other highly prevalent diseases, having a reliable prediction method is crucial. The objective of this paper is to classify peptide sequences according to their anti-angiogenic activity to understand the underlying principles via machine learning. First, the peptide sequences were converted into three types of numerical molecular descriptors based on the amino acid composition. We performed different experiments with the descriptors and merged them to obtain baseline results for the performance of the models, particularly of each molecular descriptor subset. A feature selection process was applied to reduce the dimensionality of the problem and remove noisy features – which are highly present in biological problems. After a robust machine learning experimental design under equal conditions (nested resampling, cross-validation, hyperparameter tuning and different runs), we statistically and significantly outperformed the best previously published anti-angiogenic model with a generalized linear model via coordinate descent (glmnet), achieving a mean AUC value greater than 0.96 and with an accuracy of 0.86 with 200 molecular descriptors, mixed from the three groups. A final analysis with the top-40 discriminative anti-angiogenic activity peptides is presented along with a discussion of the feature selection process and the individual importance of each molecular descriptors According to our findings, anti-angiogenic activity peptides are strongly associated with amino acid sequences SP, LSL, PF, DIT, PC, GH, RQ, QD, TC, SC, AS, CLD, ST, MF, GRE, IQ, CQ and HG.

Artificial Astrocytic Modulation of Neuron’s Output

Artificial Neuron-Glia Networks (ANGN) are feed-forward multilayer artificial networks that are composed of two types of information processing elements: one type that emulates neurons and another type that emulates astrocytes. These networks implement an astrocytic modulation that simulates the ability of astrocytes to modify the synaptic space, enhancing the weights of the connections. In this work ANGN have been implemented with a new type of astrocytic modulation observed in the brain: artificial astrocytes act on the output value of artificial neurons increasing or reducing the amount of neurotransmitter released in the synaptic terminal. For three classification problems the results of the comparison of the new type and the previous type of modulation and Artificial Neural Networks without astrocytes are shown. It is observed that depending on the problem, one or another type of astrocytic modulation is better, but in all cases the performance of artificial networks with astrocytes is superior.