Main Content Region

Ongoing Research Projects supported by Research IT

Listing of project codes and abstracts, describing work undertaken which use the resources of the compute clusters hosted by the Research IT team.

New codes can be obtained by filling in this resource application form.

Search Results
Showing 10 of 423 Results
Project Title CFD analysis of pulmonary airflow
Project Code HPC_17_00975
Principal Investigator Dr Colleen Farmer
Start Date 2017-06-02
End Date 2018-06-02
Abstract This project aims to model airflow in the lungs of a range of vertebrates to understand basic structure-function questions. The overarching objective is to expand our knowledge of the evolution of the vertebrate respiratory system.
Project Title Ion migration
Project Code HPC_17_00974
Principal Investigator Prof Khurshid Ahmad
Start Date 0000-00-00
End Date 0000-00-00
Abstract I use molecular dynamics and quantum chemistry to simulate the migration of ions in molecular channels.
Project Title In silico modelling of arterial disease
Project Code HPC_17_00973
Principal Investigator Prof Caitriona Lally
Start Date 2017-06-02
End Date 2018-06-02
Abstract Each year cardiovascular diseases such as atherosclerosis and aneurysms cause 48% of all deaths in Europe. Arteries may be regarded as fibre-reinforced materials, with the stiffer collagen fibres present in the arterial wall bearing most of the load during pressurisation. Degenerative vascular diseases such as atherosclerosis and aneurysms alter the macroscopic mechanical properties of arterial tissue and therefore change the arterial wall composition and the quality and orientation of the underlying fibrous architecture. Information on the complex fibre architecture of arterial tissues is therefore at the core of understanding the aetiology of vascular diseases. The current proposal aims to use a combination of in vivo Diffusion Tensor Magnetic Resonance Imaging, with parallel in silico modelling, to non-invasively identify differences in the fibre architecture of human carotid arteries which can be directly linked with carotid artery disease and hence used to diagnose vulnerable plaque rupture risk. Knowledge of arterial fibre patterns, and how these fibres alter in response to their mechanical environment, also provides a means of understanding remodelling of tissue engineered vessels. Therefore, in the second phase of this project, this novel imaging framework will be used to determine fibre patterns of decellularised arterial constructs in vitro with a view to directing mesenchymal stem cell growth and differentiation and creating a biologically and mechanically compatible tissue engineered vessel. In silico mechanobiological models will also be used to help identify the optimum loading environment for the vessels to encourage cell repopulation but prevent excessive intimal hyperplasia. This combination of novel in vivo, in vitro and in silico work has the potential to revolutionise approaches to early diagnosis of vascular diseases and vascular tissue engineering strategies.
Project Title Muon g-2 and the isospin breaking corrections to the hadronic contributions
Project Code HPC_17_00970
Principal Investigator Assistant Professor Marina Krstic Marinkovic
Start Date 2017-05-23
End Date 2018-05-22
Abstract Several leading lattice collaborations are investing significant efforts to reduce the uncertainty in the lattice computation of the leading hadronic contribution to the muon anomalous magnetic moment to a sub- percent level in order to verify weather the current discrepancy between the standard model estimate and the experimental measurements is a sign of new physics beyond the standard model. In order to achieve this goal, giving a precise estimate of the size of the isospin breaking effects becomes relevant. Several lattice collaborations are currently competing to estimate of this important effect and TCD researchers are part of these cutting-edge efforts. We therefore apply here for computer time to measure the isospin breaking effects of the connected part of the hadronic vacuum polarisation from the lattice by applying the recently proposed method of R123 collaboration. The decisive point of this approach is defining the leading isospin breaking effects (LIBE) of an observable, by expanding in powers of the fine structure constant alpha_em, and up and down quark mass differences, mu-md. This method is developed to enable the computation of the isospin breaking corrections order by order, and its advantage with respect to the conventional methods is that it factorises the small coefficients, which in turn amounts to relatively large signal to noise ratios. Additionally, this method is not susceptible to the systematics associated with extrapolation to the physical value of the fine structure constant. On the other hand, the fact that one has to compute a significantly larger number of quark and photon propagators than in the standard approach of quenched and full QED+QCD simulations might be considered a disadvantage. However, this can be overcome by using highly optimal inverter and careful organisation of the computation of the diagrams that is being followed in this work.
Project Title Cortical surface reconstruction from MR data in The Irish Longitudinal Study on Ageing (TILDA)
Project Code HPC_17_00969
Principal Investigator Prof Rose Kenny
Start Date 2017-05-29
End Date 2018-05-28
Abstract Cortical surface reconstruction from MR data in The Irish Longitudinal Study on Ageing (TILDA): investigating cortical structure in older Irish adults. The human cortex displays remarkable profiles of change across the lifespan, both in health and disease. A unique challenge for research is to map brain structure changes with age onto the broader physical, cognitive, social, and economic factors that may be critical to healthy, successful ageing. The Irish Longitudinal Study on Ageing (TILDA) is the largest single-cohort study of older adults in Ireland, comprising a nationally representative sample of adults aged 50+. The goal of the current project is to explore cortical change in Irish adults aged 50+, by charting relationships between cortical structure and physical and cognitive health, alongside socio-demographic and economic factors. With the assistance of TCD TCHPC architectures, T1-w MR scans collected for a subset of TILDA participants allow the cortical surface (and a range of subcortical structures) to be reconstructed with FreeSurfer software. These reconstructed cortical surfaces enable investigations of cortical thickness, area, and curvature, as they relate to age, and to the detailed physical health assays available for TILDA participants (including cardiac, optical, orthopaedic, and ambulatory health). Moreover, such cortical metrics afford the opportunity to explore relationships between brain structure and the cognitive, social, environmental, and economic circumstances of older Irish adults. This project will provide a detailed exploration of the multi-faceted changes in brain structure with age, as they relate to factors that characterise growing older in Ireland.
Project Title Theoretical study on porphyrin/ guanidine conjugates as potential binders of Guanine-Quadruplexes
Project Code HPC_17_00968
Principal Investigator Prof Isabel Rozas
Start Date 2017-04-11
End Date 2018-05-31
Abstract Prof. Rozas’s group has been working in the design, synthesis and biological study of guanidinium derivatives as DNA minor groove binders during the last 15 years and these finding reveals very promising derivatives with activity as cell growth inhibitors.1-3 Continuing with our interest in the interactions established between guanidinium derivatives and Nucleic Acid structures, we have designed novel N-substituted porphyrins linked to DNA-binding small molecules such as guanidine-like systems to yield new dual-modality portmanteau binders to achieve selective GQ stabilization and enhanced detection. Porphyrins are an important class of compounds that has gained significant recognition as DNA binding agents. In particular, 5,10,15,20-tetrakis(N-methylpyridyl)porphyrins were found to bind to GQs through π–π-stacking but suffer from poor selectivity towards GQs.4,5 In context of interesting GQ binding properties of porphyrins and our onging work in the area of guanidines as DNA-minor groove binding agents, we propose now to study the conformation of a number of porphyrin/guanidine derivative conjugates and their potential interaction with known guanine quadruplexes structures, already reported in the RCSB PDB.6,7 This study will allow us to better understand the orientation of our compounds in the domain of GQ and important binding interactions with GQ. Moreover, the study also helps us to design new GQ-ligands in order to improve the GQ binding activity and selectivity over dsDNA.
Project Title Light field image processing
Project Code HPC_17_00967
Principal Investigator Prof Aljosa Smolic
Start Date 2017-04-10
End Date 2017-11-30
Abstract Light field imaging has been recently gaining a lot of attention in the image processing community, and as such a particular topic of interest in the V-SENSE project, lead by Prof. Aljosa Smolic. A light field captures light rays coming from different angles, and can be represented by a collection of 2D images arranged on a 2D grid, hence forming 4D visual data. Applications include post-capture refocusing, depth estimation, virtual viewpoint rendering, de-noising, super-resolution. Due to their very nature, processing light fields can require large memory amounts and high computational power. In this project we aim to investigate and evaluate novel methods for light field enhancement, including de-noising and super-resolution, as well as depth estimation.
Project Title Frustrating Chemistry – Towards a Quantitative Model for Lewis Pair Association and Reactivity
Project Code HPC_17_00966
Principal Investigator Dr Carl Poree
Start Date 2017-04-10
End Date 2018-12-24
Abstract The formation of covalently-bound adducts between Lewis acids and Lewis bases has long been known – the archetype (ammonia borane, NH3∙BH3) was first prepared in the 1920s,[1] and its structure established in 1955.[2] These adducts are generally relatively unreactive, in marked contrast to their components. However, in recent years, the existence of so-called frustrated Lewis pairs (FLPs) – in which adduct formation does not occur – has been noted. This would arguably be of only marginal interest, if it were not for the remarkable ability of FLPs to activate relatively inert molecules such as hydrogen (H2) and carbon dioxide (CO2).[3] Previously, such modes of reactivity were available only to (often expensive) transition metal-based catalysts. Despite this considerable and growing interest in the field, a key, fundamental deficiency remains. The definition of a Lewis pair as ‘frustrated’ is primarily based on structure, and the inability of a given pair of Lewis acids and bases to form an adduct is generally considered to be due to the steric bulk of one or both partners. However, neither clear structural criteria nor guiding principles for the design of FLPs have emerged to date. Given the challenges inherent in handling the components required to form an FLP – they are often air- and moisture-sensitive – the design of more robust catalysts is of considerable interest. The prevailing mechanistic paradigm for H2 activation by FLPs is formally termolecular (i.e. requiring the encounter of three molecules in the rate-determining transition state), which is statistically highly improbable, absent the formation of a presumably weakly-bound bimolecular acid-base encounter complex. Computational studies to date have been predicated on the formation of such an encounter complex, but have not addressed the general questions of structure relevant to the formation of such species. Limited spectroscopic evidence for weak Lewis acid – Lewis base interactions in solution has also been reported, but is limited to two FLP systems (either tBu3P or Mes3P with B(C6F5)3).[4] The aim of the proposed work is to develop a model for the formation of traditional Lewis adducts or FLPs from pairwise combinations of phosphines and boranes, and to evaluate their abilities to activate small molecules (in the first instance, H2). This will involve quantum chemical calculations (predominantly employing density functional theory methods) in combination with statistical analysis with the overriding aim of determining key structure-activity relationships, allowing for the informed design of improved catalysts. In the longer term, experimental validation of the predictions generated by this study is envisaged. [1] A. Stock, E. Pohland, Ber. Dtsch. Chem. Ges. A/B 1925, 58, 657-661. [2] S. G. Shore, R. W. Parry, J. Am. Chem. Soc. 1955, 77, 6084-6085. [3] D. Stephan, G. Erker, Angew. Chem. Int. Ed. 2015, 54, 6400-6441 and references therein. [4] L. Rocchigiani, G. Ciancaleoni, C. Zuccaccia, A. Macchioni, J. Am. Chem. Soc. 2014, 136, 112-115.
Project Title Reproducible prediction for large imaging data
Project Code HPC_17_00965
Principal Investigator Dr Robert Whelan
Start Date 2017-03-06
End Date 2018-03-05
Abstract Overestimation of effect sizes and lack of reproducibility is a problem endemic to all medical and psychological research (Ioannidis, 2005a/b, 2008; Open Science Collaboration, 2015; Ware & Munafo, 2014), and has become increasingly addressed in the field of neuroimaging in recent years (see Pernet & Poline, 2015; Poldrack & Poline, 2015; Ioannidis et al., 2014). With the move toward Big Data in neuroscience the field has taken an important step towards addressing this reproducibility crisis. However, the availability of large imaging datasets also calls for the use of truly robust analysis methods (see also Fritsch et al., 2015). In order for neuroimaging research to advance clinical practice and scientific inquiry it is therefore necessary to adopt truth-supporting analysis approaches developed for data science (Iniesta, Stahl, & McGuffin, 2016; Yarkoni & Westfall, n.p.; Jollans & Whelan, 2016). Machine Learning algorithms such as Support Vector Machines, Random Forests, or Naïve Bayes classifiers have become standard tools for regression and classification analyses in the neuroimaging community. However, the majority of these tools were not developed with imaging data in mind, and neither the ability of these methods to deal with the inherent small effect sizes and multicollinearity of imaging data, nor how data dimensionality affects performance of these algorithms has been established in the neuroimaging literature. In this project we intend to empirically evaluated the efficacy of some of these approaches for 2nd level between-subject regression analysis with high-dimensional neuroimaging data, and how sample size and number of input variables (i.e. features) interact with elements of machine learning protocols.
Project Title Omnidirectional video coding
Project Code HPC_17_00964
Principal Investigator Prof Aljosa Smolic
Start Date 2017-03-03
End Date 2018-05-11
Abstract 360-degree video is attracting an increasing amount of attention in the context of Virtual Reality (VR). Owing to its very high-resolution requirements, existing professional streaming services for 360-degree video suffer from severe drawbacks. This research introduces a novel method to encode 8K resolution 360-degree video and provide an enhanced VR experience using Head Mounted Displays (HMDs).