1731 research outputs found
Sort by
Computational Systems Biology Applied To Human Metabolism. Mathematical Modelling and Network Analysis.
Full text linkHuman metabolism, an essential and highly organized process, which is required to run and maintain cellular processes and to respond to shifts in external and internal conditions, can be described as a complex and interconnected network of metabolic pathways.
Computational systems biology provides a suitable framework to study the mechanisms and interactions of this network and to address questions that are difficult to reproduce in vitro or in vivo.
This dissertation contributes to the development of computational strategies which help to investigate aspects of human metabolism and metabolic-related disorders.
In the first part, we introduce mathematical models of folate-mediated one-carbon metabolism in the cytoplasm and subsequently in the nucleus.
A hybrid-stochastic framework is applied to investigate the behavior and stability of the complete metabolic network in response to genetic and nutritional factors.
We analyse the effect of a common polymorphism of MTHFR, B12 and folate deficiency, as well as the role of the 5-formyltetrahydrofolate futile cycle on network dynamics.
Furthermore, we study the impact of multienzyme complex formation and substrate channelling, which are key aspects related to nuclear folate-mediated one-carbon metabolism.
Model simulations of the nuclear model highlight the importance of these two factors for normal functioning of the network and further identify folate status and enzyme levels as important influence factors for network dynamics.
In the second part, we focus on metabolic syndrome, a highly prevalent cluster of metabolic disorders.
We develop a computational workflow based on network analysis to characterise underlying molecular mechanisms of the disorder and to explore possible novel therapeutic strategies by means of drug repurposing.
To this end, genetic data, text mining results, drug expression profiles and drug target information are integrated in the setting of tissue-specific background networks and a proximity score based on topological distance and functional similarity measurements is defined to identify potential new therapeutic applications of already approved drugs. A filtering and prioritization analysis allow us to identify ibrutinib, an inhibitor of bruton tyrosine kinase, as the most promising repurposing candidate
Learning in Low Data Regimes for Image and Video Understanding
Full text linkThe use of Deep Neural Networks with their increased representational power has allowed for great progress in core areas of computer vision, and in their applications to our day-to-day life. Unfortunately the performance of these systems rests on the "big data" assumption, where large quantities of annotated data are freely and legally available for use. This assumption may not hold due to a variety of factors: legal restrictions, difficulty in gathering samples, expense of annotations, hindering the broad applicability of deep learning methods.
This thesis studies and provides solutions for different types of data scarcity: (i) the annotation task is prohibitively expensive, (ii) the gathered data is in a long tail distribution, (iii) data storage is restricted.
For the first case, specifically for use in video understanding tasks, we have developed a class agnostic, unsupervised spatio-temporal proposal system learned in a transductive manner, and a more precise pixel-level unsupervised graph based video segmentation method. At the same time, we have developed a cycled, generative, unsupervised depth estimation system that can be further used in image understanding tasks, avoiding the use of expensive depth map annotations.
Further, for use in cases where the gathered data is scarce we have developed two few-shot image classification systems: a method that makes use of category-specific 3D models to generate novel samples, and one that increases novel sample diversity by making use of textual data.
Finally, data collection and annotation can be legally restricted, significantly impacting the function of lifelong learning systems. To overcome catastrophic forgetting exacerbated by data storage limitations, we have developed a deep generative memory network that functions in a strictly class incremental setup
Modeling and communicating the dynamics of energy market
No full textThe focus of my Ph.D. project is the investigation of the analogies between the dynamics of oil production, the economy and the physical laws that are related to every natural process. The difficulty to retrieve data on the oil market has suggested me that there is a correspondence in Science Communication: indeed, while in the magazines we find many articles that talk about climate change, for instance, we seldom met articles on related topics with keywords such as "energy resources" or "energy depletion".
I started to investigate the relationship between the phase-plot of worldwide oil production (1965-2014) and its price. My idea was that the laws regulating the convective intensity and the oil production on one side, and the energy flow in human society on the other, were similar. In particular, this analogy could be true for the change from a slow diffusion process to a faster convective movement.
In a first approximation, I tried to describe the phase-plot system in a theoretical way: I imagined the system without perturbations with a worldwide average EROI that varies from 60 (1965) to 15 (2014). Similarly, I picked up the value of world production in 1965 (1,567 Million ton/years) and the same value in 2014 (4,220 Million ton/years) and I assumed a constant increment of the production. Obviously, the phase-plot de-scribes the evolution of the real system and appropriately shows two evident "loops" in correspondence of the two major oil crises in recent times (in the '70s of the last century and in the '10s of the current century), although it is also apparent a "background" (or floor) price the system has never overcome.
Following the idea of a possible description of the global oil market trend as a dynamical system, I investi-gated the eventual analogies between oil market and a classical dynamical system which is well known to exhibit an almost doubly cyclic behavior, i.e., the Lorentz attractor.
However, this suggestion remained nothing better than that, because the major criticism is that it is rather difficult to conceive that only two loops in a phase-plot could be assimilated tout court to a Lorenz attractor: the investigated period 1965-2014 is in fact too short to validate such an analogy. Therefore, the output of the research was an article deposited it in the arXiv online repository .
The relationship between natural phenomena and price-production dynamics of oil extraction, however, could be seen from another point of view. Indeed, the phase-plot, behind the "random walk" initially de-scribed as a Lorenz attractor, suggested us two peaks in correspondence with the upper part of those loops. This "swinging" (macro)behavior is rather similar to that of the theoretical model where we only have two stocks of oil and two prices for them: the first delimited by a price of 100 for a production in the range 1,000-1,500 barrels/day. In this ideal case, following the consumption, we expect the same swinging behavior: if the consumption is in the range between 0 and 1,000 barrels/day, the oil at the lower price will be only used, with a price (ideally) of 100 ). In this ideal dynamics, the price suddenly jumps to the higher level instead of slowly shifting up. If the oil price increases, however, the consumption decreases and, sooner or later, the society comes back to the previous range of extraction intensity. The main characteristic of the simple oil-price dynamics here de-scribed is that there are two rapid movements on the cycle (rise and descent of the price) and two slow movements (consumption that goes up and down, to adapt to the oil price). At this level could be useful in terms of EROI instead of oil stocks (with high and low prices). Therefore, natural variables seem to be more similar to medium price and medium EROI (fast the former, slow the latter).
The Lotka-Volterra equations describe the prey-predator ecological mechanisms and many cases are well known and well-studied. One of them seems to have the same features of the simple oil market behavior just described. More specifically, the analogy is between a prey, the American spruce, and a predator, the caterpillars of the species Choristoneura fumiferana that feed on the spruce. The caterpillar population is regarded as the fast variable, since there are periodically observed demographic outbreaks of this species, considered a real scourge , whereas the spruce leaf whole surface is assumed to be the slow variable be-cause the regeneration of the leaves - and not only - is a process that lasts several decades.
This dynamic model requires a good evaluation of the worldwide EROI for the oil. This is, in any case, fun-damental as an index of the energy quality used by the society. This is the reason why the EROI is crucial to determine the goodness of an energy resource in general and particularly for oil and gas, that nowadays sat-isfy about 57% of primary energy demand .
This dimensionless index is generally defined as the ratio between the energy extracted from a given re-source and the energy costs sustained to get that energy. In a following step of my research, I tried to set up an alternative method for the calculation of the EROI of oil companies. The difficulties to retrieve the data by the oil companies is notorious, thus the strategy consisted in using as a proxy of the energy costs, i.e. the available data about the CO2 emissions of the same oil companies, as reported in their own sustainability reports (SRs). International organizations such as IPCC and WBCSD recommend to the involved companies to compile these reports, but they are not mandatory. The second step was to use, as a proxy of the energy ex-tracted, the CO2 emissions estimate obtained by a stoichiometric conversion of the oil production declared by the oil companies.
The resulting estimates of EROI are rather homogeneous and not too different from the values reported in the literature. The method could be suitable for year-by-year comparison of the time evolution of this im-portant energy quality parameter for the individual energy-producing and energy-delivering companies . I defined this parameter as the "corporate's EROI".
In particular, this last work had its main difficulty in finding data to make comparisons between the different oil companies. As mentioned above, this is also reflected in the communication sector. If we look at the fol-lowing diagram, we can discover, for example, the differences between related argument ("climate change") and keywords like "energy resource" or "energy depletion". The graph includes all the terms worldwide for the last year (October 1st, 2017 - October 1st, 2018) . I am personally committed to filling the gap (thanks to my previous background in Science Communication), to sensitize the citizenship to the energy transition problem and last year, at the Bright event (European Researchers’ Night) I have developed a game about the Hubbert oil peak to play with people and explain to them the dynamics of peak and resource depletion
Regular black hole and cosmological spacetimes in Non-Polynomial Gravity theories
Full text linkGeneral Relativity is known to suffer from singularities at short distances, which indicates the breakdown of its predictability, for instance at the center of black holes, and in the very early universe. This is one of the main reason to look for a Quantum Theory of Gravity, that would describe spacetime geometry as a quantum field, and possibly cure these classical singularities. However, no consensus on the topic has yet been reached, as many different approaches have been proposed, but none has yet received an experimental confirmation. This is in part due to the extraordinary small scale at which quantum gravitational effects are expected to become dominant, and to the technical difficulty to make unambiguous predictions. For this reason, many works have focused on the so-called effective approaches in which the possible high energy corrections to General Relativity are classified, and their theoretical and ob- servational predictions derived, with the idea that among these modifications, some could come as the semi-classical limits of quantum gravity theories. A way to discriminate between the different proposals is precisely the absence of singular geometries in their solutions. In the first two Chapters of this thesis, we will present such an effective approach, in which the action of General Relativity is modified at high energy by non-polynomial curvature invariants, which are constructed in such a way that the dynamical spherically symmetric sector of these theories (which contain both cosmological and non-rotating black hole spacetimes) yield second order field equations. These properties of the non-polynomial invariants follow from a peculiar algebraic identity satisfied by the Cotton tensor in this class of geometries. As we will see in the last two Chapters, having second order dynamical spherically symmetric field equations is necessary in order to recover some quantum corrected geometries that have been found from more fundamental approaches like Loop Quantum Cosmology and Asymptotic Safety, within its Einstein-Hilbert truncation. The existence of such gravitational models provides an interpretation of two-dimensional Horn- deski theory as describing the dynamical spherically symmetric sector of specific higher dimensional non-polynomial gravity theories. Therefore, it allows to have some concrete d-dimensional formu- lations of the two-dimensional Einstein-Dilaton and Lovelock Designer effective approaches that have been studied extensively, in particular to find and study the properties of non-singular black holes. This enables us to propose two four-dimensional effective-like actions, which are constructed in such a way that their dynamical spherically symmetric sectors decompose in the same way as those of General Relativity and Gauss-Bonnet gravity. In the remaining Chapters, we essentially investigate the solutions and properties of these theories. It is shown that the first one leads to regular (A)dS-core black hole solutions, with the correct quantum correction to their Newton potentials and logarithmic correction to their entropies. The charged generalization is considered, and a way to avoid the mass inflation instability of their inner horizons is found, provided that a bound between the mass and the charge is satisfied. In Chap. 4, we establish a reconstruction procedure able to find theories admitting as solutions the Modesto semi-polymeric black hole, as well as the D’Ambrosio-Rovelli and Visser-Hochberg geometries. All these black holes are regular and derived or inspired by quantum gravity results. They have many properties in common, as for example the fact that they automatically regularize the Coulomb singularity of a static electric field. Finally, the last Chapter is devoted to the theory whose dynamical spherically symmetric sector is a generalization of the one of Gauss-Bonnet gravity. It is shown that the Loop quantum cosmology bounce universe and some Asymptotic Safety black holes can be reconstructed from two members of these theories. In particular, the associated black hole solutions of the first are regular, and the associated cosmological solution of the second is as well, and describe a universe which is eternal in the past, and behaves as de Sitter spacetime in the limit of infinite past. Some generalizations of these results are provided, and the Mimetic gravity formulations of the cosmological solutions are found
Development of Devices Based on Electrically Actuated Soft Elastomers
Full text linkDielectric elastomer (DE) actuators are electromechanical transducers that essentially consist of one layer of an insulating soft elastomer coated on both sides with com- pliant electrodes. When a voltage is applied between the electrodes, an electrostatic pressure deforms the elastomer triggering the motion of the actuator. In this the- sis, this principle is exploited for the development of three different actuators: an electroactive compression bandage, a hydrostatically coupled actuator for use in the field of soft manipulators and a dielectric elastomer based inchworm-like robot able to perform locomotion. By doing so, several challenges related to the design, to the modeling and to the manufacturing of this kind of devices are raised and tackled. During the development of the electroactive compression bandage, the issue of electrical insulation and prevention of electrical discharge in wearable devices was addressed by using coating layers as an interface between the DE actuator and the human body. Both experimental investigations and a finite electro-elasticity analyti- cal model showed that the passive layers play a key role for an effective transmission of the actuation from the active layers to the load. Indeed, the model showed that by increasing the number of electroactive layers, the pressure variation can be increased, although with a saturation trend, providing a useful indication for future designs of such bandages. The second piece of work here reported consists in a design upgrade of the Hy- drostatically Coupled Dielectric Elastomer Actuator (HC-DEA), already known in the literature, that enable its use in the field of soft manipulators. The new design fea- tures segmented electrodes, which stand as four independent elements on the active membrane of the actuator, enabling it for generating both out of plane and in plane motions. This novel design makes the actuator suitable for delicate transportation of a flat object. This capability was proven via an experimental investigation in which a flat Petri dish was roto-translated on a platform composed of two actuators. The electromechanical transduction performance of the actuator was characterized and its contact mechanics was modeled.
Finally, a smart robot structure that exploits anisotropic friction to achieve stick- slip locomotion is presented. The robot, which is made out just of a plastic beam, a planar dielectric elastomer actuator and four bristle pads with asymmetric rigid metallic bristles, exploits the resonance condition to reach the maximum locomotion speed. The fundamental frequency of the structure, which was estimated both ana- lytically and numerically, was identified within the range of frequencies in which the top locomotion speed was observed during the experiments to be identified
Implementing evidence-based treatments for developmental dyslexia: a comparison between different approaches
Full text linkDevelopmental Dyslexia (DD) is one of the most common neurodevelopmental disorders across cultures. Children affected by DD struggle to read fluently and/or correctly, despite normal intelligence, the absence of other psychological or neurological symptoms, and standard reading education. Not only does this condition affect academic achievement, but it is also associated with a number of other negative consequences across the lifespan, such as
an enhanced risk of psychological distress and mental health problems. Despite considerable efforts to identify the underlying cause of dyslexia, agreement on a single interpretation has not yet been reached. DD is commonly described as a languagerelated disorder, with compromised phonological abilities being considered as the core
deficit. Nonetheless, a growing body of evidence supports multifactorial models: reading is a complex cognitive process, involving not only phonological skills, but also auditory sensory processes, visual-spatial abilities, attention and memory. In this regard, various studies have
documented a relationship between dyslexia and deficits in the Executive Functions (hereafter EFs), which can be defined as a cluster of general-purpose control mechanisms that modulate various cognitive sub-processes. However, the relationship between cognitive correlates and reading impairments is still a controversial issue.
The present project aims to analyse the fundamental characteristics and the efficacy of different rehabilitation methods for Developmental Dyslexia. Although the neurocognitive causes of DD are still hotly debated, researchers agree that the main challenge is the
remediation, that is, how to improve children’s reading fluency and accuracy. The most common approach has been to devise sophisticated remediation programs that train
sub-skills of reading, especially phonological skills and auditory perception. Despite the promising results, the improvements in these sub-skills do not automatically transfer in better reading abilities in all subjects (especially regarding reading fluency), thus giving rise to the issue related to "non-responders" or "poor responders".
Since the present data gives firm indications of the need to individualize intervention based on neuropsychological testing, the aim of this project is to investigate the efficacy of new types of treatment based on a multifactorial, probabilistic, model of the disorder.
Consequently, this project consisted of two parts: specifically, in the first study, we compared phonological-based treatment with computerized cognitive training of the executive functions (e.g., attention, working memory, planning, inhibition). The results of this study clearly
pointed out an advantage both in terms of improvements in EFs and literacy skills for the group who undertook the Integrated training, i.e., the group that underwent 12 hours of Cognitive training prior to 12 hours of Phonological-based treatment. Next, the second study aimed to explore the efficacy of a video game Skies of Manawak purposefully designed to train several EFs. Indeed, it has been showed that the existing treatments are not sufficiently captivating and motivating and, thus, we developed a tool ex
novo in order to obtain overall improvement and higher chances of transfer to untrained tasks. Our goal was to investigate whether playing this video game may enhance EFs following intervention, and whether these improvements transfer to important literacy skills in typically
developing (Study 2 – Part A) and dyslexic children (Study 2 – Part B). All children underwent 12 hours of training, distributed over 6 weeks, either on Skies of Manawak or on a
control computerized activity (Scratch). Assessments upon training completion indicated greater improvements in executive functioning and reading efficiency after Skies of
Manawak than after the control training in both studies. Interestingly, the advantage in reading skills was maintained in a follow-up test 6 months later and seemed to generalize to academic performance (i.e., Italian marks).
Overall findings highlighted promising effects of the training programs on children’s cognition, making way for future studies investigating the underlying brain mechanisms and the factors leading to treatment success
La prosa latino-francese di argomento troiano del codice Barb. lat. 3953 e la fortuna medievale della materia troiana in Italia
Full text linkLa tesi offre una proposta di edizione critica e un’analisi della prosa di argomento troiano trasmessa dal codice Città del Vaticano, Biblioteca Apostolica Vaticana, Barberiniano latino 3953. L’opera, in latino, ma con inserti in antico-francese, viene indagata e contestualizzata sotto l’aspetto filologico e critico-letterario. Il lavoro è completato da uno studio introduttivo sulla fortuna della materia troiana in Italia tra i secoli XIII-XV, nelle sue manifestazioni letterarie (in lingua oitanica e nei volgari italo-romanzi) e artistiche. In appendice è fornita una lista provvisoria di testimoni dell’"Historia destructionis Troiae" di Guido delle Colonne
Personajes divergentes y motivos caballerescos: el caso de Lidamor de Escocia (1534), Valerián de Hungría (1540), Philesbián de Candaria (1542) y Cirongilio de Tracia (1545).
Full text linkEl presente trabajo de investigación tiene por objeto el estudio de los personajes divergentes en el marco de unos libros de caballerías del siglo XVI que no conforman ningún ciclo, ni vuelven a editarse durante el período de apogeo del género. Se trata, pues, de obras que permanecen aisladas frente a los ciclos de mayor éxito, cuyos textos dan lugar, en cambio, a distintas ediciones a lo largo de la centuria. Los libros de caballerías que conforman el corpus del presente trabajo son Lidamor de Escocia de Juan de Córdoba (1534), Valerián de Hungría de Dionís Clemente (1540), el anónimo Philesbián de Candaria (1542) y Cirongilio de Tracia de Bernardo de Vargas (1545). Posiblemente debido al escaso éxito del que gozaron, estas obras no fueron objeto de estudios profundizados por parte de la crítica y muchos de sus aspectos estilísticos y temáticos quedan por explorar.
En las inmensas tramas de los libros de caballerías se alternan distintos personajes y al lado de hermosas damas y valientes caballeros se encuentran también enanos cobardes, salvajes velludos, gigantes horroríficos y magos (casi) todopoderosos. En el marco del presente trabajo se presta atención a algunas magas y algunos gigantes que intervienen en las tramas de las obras elegidas. La configuración de estos personajes se da conforme al canon bajo unos aspectos, a la vez que, en ocasiones, es el fruto de interesantes desvíos.
El enfoque teórico adoptado para la investigación acerca de los personajes divergentes privilegia el motivo literario como elemento fundamental de la estructura de las obras. En efecto, los libros de caballerías que se difunden en España durante el Renacimiento son herederos de la literatura medieval y los motivos de que esta se nutre pasan, en cuanto matériel roulant, de una tradición a otra. El estudio de los tipos y los motivos folclóricos que empieza a principios del siglo XX gracias a la labor de Aarne y Thompson, se ha demostrado rentable para las investigaciones en el ámbito de la épica y otros géneros de la literatura hispánica. La creación, a lo largo del siglo pasado, de varios índices dedicados a corpora literarios diferentes da buena muestra del éxito del tal enfoque. Al lado de la catalogación del material folclórico, se ha ido concretando la idea de realizar un índice ad hoc para los libros de caballerías del siglo XVI basado en un sistema proprio. Así pues, el Índice y estudio de motivos de los libros de caballerías castellanos (1508-1516) de Bueno Serrano (2008) se configura como un instrumento de localización de las unidades narrativas caballerescas y constituye el punto de partida para llevar a cabo el análisis crítico de los personajes divergentes.
Magos y gigantes se presentan en una fichas donde se marca su trayecto narrativo y se anotan las matizaciones identificadas en el uso de los motivos respecto a su realización en los seis primeros libros de caballerías analizados por Bueno Serrano. Las oscilaciones se dan en la interacción entre el nivel paradigmático y, pues, teórico del motivo y su realización en el nivel sintagmático, es decir, en cada nuevo contexto. Para hacer hincapié en la importancia que cobra esta interacción a la hora de utilizar el índice como instrumento crítico, se asocia a cada motivo el fragmento textual en el que este se realiza.
Los datos recogidos mediante el cotejo entre las obras insertas en el corpus y los textos que conforman el paradigma caballeresco dan constancia de que los libros de caballerías no son ni mucho menos todos iguales, sino que los autores reelaboran a un material conocido para amoldarlo en el horizonte de sus propias obras. Desde esta perspectiva, el estudio de las divergencias que se aprecian en el nivel de los motivos puede ir marcando las pautas de evolución del género en el período de su gran difusión. Debido a ello es necesario aplicar el análisis sistemático de las unidades narrativas a un corpus más amplio de obras caballerescas, con el fin de poner de relieve su desarrollo. La futuras investigaciones en este ámbito pueden ser facilitadas por la cooperación de las Humanidades Digitales mediante la creación de una base de datos donde se coleccionen los personajes divergentes, los motivos que desencadenan y las posibilidades de matización de los mismos
Thermomechanical modelling of powder compaction and sintering
Full text linkAn elastic-visco-plastic thermomechanical model for cold forming of ceramic powders and subsequent sintering is introduced and based on micromechanical modelling of the compaction process of granulates. Micromechanics is shown to yield an upper-bound estimate to the compaction curve of a granular material, which compares well with other models and finite element simulations. The parameters of the thermomechanical model are determined on the basis of available data and dilatometer experiments. Finally, after computer implementation, validation of the model is performed with a specially designed ceramic piece showing zones of different density. The mechanical model is found to accurately describe forming and sintering of stoneware ceramics and can therefore be used to analyze and optimize industrial processes involving compaction of powders and subsequent firing of the greens
A phylogenetic framework for large-scale analysis of microbial communities
Full text linkThe human microbiome represents the community of archaea, bacteria, micro-eukaryotes, and viruses present in and on the human body. Metagenomics is the most recent and advanced tool that allows the study of the microbiome at high resolution by sequencing the whole genetic content of a biological sample. The computational side of the metagenomic pipeline is recognized as the most challenging one as it needs to process large amounts of data coming from next-generation sequencing technologies to obtain accurate profiles of the microbiomes. Among all the analyses that can be performed, phylogenetics allows researchers to study microbial evolution, resolve strain-level relationships between microbes, and also taxonomically place and characterize novel and unknown microbial genomes. This thesis presents a novel computational phylogenetic approach implemented during my doctoral studies. The aims of the work range from the high-quality visualization of large phylogenies to the reconstruction of phylogenetic trees at unprecedented scale and resolution. Large-scale and accurate phylogeny reconstruction is crucial in tracking species at strain-level resolution across samples and phylogenetically characterizing unknown microbes by placing their genomes reconstructed via metagenomic assembly into a large reference phylogeny. The proposed computational phylogenetic framework has been used in several different metagenomic analyses, improving our understanding of the complexity of microbial communities. It proved, for example, to be crucial in the detection of vertical transmission events from mothers to infants and for the placement of thousands of unknown metagenome-reconstructed genomes leading to the definition of many new candidate species. This poses the basis for large-scale and more accurate analysis of the microbiome