In this thesis various analyses and modeling of asphalt mixtures are represented. It is composed of four main parts. In the first part we build up the database of individual measurements of bituminous binder, aggregate and asphalt mixtures. These data are analyzed using statistical methods which determine whether the effects of factors and their interactions are statistically significant. In addition we have used artificial neural networks for modeling relationship between different parameters and air voids content in asphalt mixtures and voids content in a mixture of aggregate. In the second part of the study cyclic triaxial tests and wheel tracking tests were performed to determine the resistance of asphalt mixtures to permanent deformation at high temperatures. Furthermore other conventional analyses of asphalt mixtures were performed to use them in different models. Experiments were done for all principal asphalt mixtures to obtain an extensive model of asphalt behaviour for mixtures which contain the most and the least air voids. Various linear models, their validation and extensions to other asphalt mixtures are featured. The third part deals with the use of different artificial neural networks for modeling permanent deformation from triaxial tests. The last part presents the use of ultrasonic shear wave reflection (USWR) method on bituminous binder. For the evaluation of this method three paving grade bitumens and two polymer modified bitumens were tested. Differences in the evolution of shear wave reflection coefficient with temperature in the case of different bitumen types indicate that the presented ultrasonic wave reflection method could represent an advanced non-destructive technique for monitoring the hardening process of different types of bitumen.
D.09 Tutoring for postgraduate students
COBISS.SI-ID: 7096929We briefly review the history of the connectivity index from 1975 to date. We hope to throw some light on why this unique, by its design, graph theoretical molecular descriptor continues to be of interest in QSAR, having wide use in applications in structure–property and structure–activity studies. We will elaborate on its generalizations and the insights it offered on applications in Multiple Regression Analysis (MRA). Going beyond the connectivity index we will outline several related developments in the development of molecular descriptors used in MRA, including molecular ID numbers (1986), the variable connectivity index (1991), orthogonal regression (1991), irrelevance of co-linearity of descriptors (1997), anti-connectivity (2006), and high discriminatory descriptors characterizing molecular similarity (2015). We will comment on beauty in QSAR and recent progress in searching for similarity of DNA, proteins and the proteome. This review reports on several results which are little known to the structure–property–activity community, the significance of which may surprise those unfamiliar with the application of discrete mathematics to chemistry. It tells the reader many unknown stories about the connectivity index, which may help the reader to better understand the meaning of this index. Readers are not required to be familiar with graph theory.
B.04 Guest lecture
COBISS.SI-ID: 5739802Rok Martinčič completed his doctoral work in the field of molecular and QSAR modeling in order to design new antioxidants and to clarify their passage across the cell membrane. The models built represent modern, useful tool for further studies of anti-oxidants and their passage through the cell membrane, while the proposed approaches in the field of QSAR modeling significantly contribute to the development of models of high quality. QSAR model was developed with structural descriptors of known antioxidant compounds. It is capable of predicting antioxidant potency of new compounds. Rok Martinčič has experimentally confirmed antioxidant activity for nearly half of the tested compounds, which had been predicted as active. He also confirmed the hypothesis that the fragment, which is common to pulvinic acid and coumarin, is the source of antioxidant activity, and the substituents attached to it are responsible for increasing or reducing activities. The compounds that showed activity in the experimental tests under various oxidative conditions, and are capable of crossing the cell membrane, represent a promising starting point for the development of a clinically useful antioxidants.
D.09 Tutoring for postgraduate students
COBISS.SI-ID: 281987840In the framework of the doctoral dissertation, Jure Borišek designed and experimentally evaluated the inhibitors of cathepsin K (Cat K) and autolysin E (AtlE). In the first part of his work Jure designed a new series of Cat K inhibitors based on N-(functionalized benzoyl)-homocycloleucyl-glycinonitrile scaffold using in silico approach. Synthesized compounds were highly selective for Cat K when compared with cathepsins L and S with the Ki values in the 10-30 nanomolar range. The kinetic studies revealed that the new compounds exhibited reversible tight binding to Cat K, while the X-ray structural studies showed covalent and non-covalent binding between the nitrile group and the catalytic cysteine Cys25 site. The second part of his research was dedicated to development of novel potential AtlE inhibitors of bacterial strain Staphylococcus aureus (S. aureus). Using computer-aided structural drug design methods and pharmacophore modeling, he managed to discover a novel chemical class representing potential AtlE inhibitors, which had binding affinities in the low micromolar range.
D.09 Tutoring for postgraduate students
COBISS.SI-ID: 282440704The awarded posted was presented at the CMTPI2015 conference. Tjaša Tibaut presented an overview and application of use of free or commercial internet resources for prediction of potential enzyme binding sites. The model study included the binding sites of autolyzin, enzyme from the hydrolazes family, which degrades the cell wall membrane.
E.02 International awards
COBISS.SI-ID: 5740058