Jožef Stefan Institute

About: Jožef Stefan Institute is a facility organization based out in Ljubljana, Slovenia. It is known for research contribution in the topics: Liquid crystal & Dielectric. The organization has 3828 authors who have published 12614 publications receiving 291025 citations.

Evolutionary Dynamics and Evolutionary History in the RTE Clade of Non-LTR Retrotransposons

Abstract: This study examined the evolutionary dynamics of Bov-B LINEs in vertebrates and the evolution of the RTE clade of non-LTR retrotransposons. The first full-length reptilian Bov-B LINE element is described; it is 3.2 kb in length, with a structural organization typical of the RTE clade of non-LTR retrotransposons. The long-term evolution of Bov-B LINEs was studied in 10 species of Squamata by analysis of a PCR-amplified 1.8-kb fragment encoding part of apurinic/apyrimidinic endonuclease, the intervening domain, and the palm/fingers subdomain of reverse transcriptase. A very high level of conservation in Squamata Bov-B long interspersed nuclear elements has been found, reaching 86% identity in the nearly 600 amino acids of ORF2. The same level of conservation exists between the ancestral snake lineage and Ruminantia. Such a high level is exceptional when compared with the level of conservation observed in nuclear and mitochondrial proteins and in other transposable elements. The RTE clade has been found to be much more widely distributed than previously thought, and novel representatives have been discovered in plants, brown algae, annelids, crustaceans, mollusks, echinoderms, and teleost fishes. Evolutionary relationships in the RTE clade were deduced at the amino acid level from three separate regions of ORF2. By using different independent methods, including the divergence-versus-age analysis, several examples of horizontal transfer in the RTE clade were recognized, with important implications for the existence of HT in non-LTR retrotransposons.

Lysosomal enzymes, cathepsins in brain tumour invasion

Abstract: The expression patterns of different classes of peptidases in central nervous system (CNS) tumours have been most extensively studied in astrocytomas and meningiomas. Although the two types of tumours are very different in most respects, both may invade locally into normal brain. This process of invasion includes increased synthesis and secretion of lysosomal proteolytic enzymes – cathepsins. Aspartic endopeptidase cathepsin (Cat) D levels were found to be elevated in high-grade astrocytoma and partial inhibition of glioblastoma cell invasion by anti-Cat D antibody suggests that the enzyme activity is involved in the invasion process. Several studies on cysteine endopeptidase (CP) Cat B in gliomas agreed that transcript abundance, protein level and activity of Cat B increased in high-grade astrocytoma cultures compared with low-grade astrocytoma cultures and normal brain. Moreover, in glioma biopsies Cat B levels correlated with evidence of clinical invasion and it has been demonstrated that Cat B both in tumour cells and in endothelial cells can serve as a new biological marker for prognosis in glioblastoma patients. A high level of Cat B protein was also a diagnostic marker for invasive types of meningioma, distinguishing between histomorphologically benign, but invasive meningiomas and noninvasive, so-called clear–benign meningiomas. Cat L was also significantly increased in high-grade astrocytoma compared with low-grade astrocytoma and normal brain. Specific Cat L antibodies and antisense Cat L RNA transfection significantly lowered glioblastoma cell invasion. In meningioma, Cat L was a less-significant marker of invasion than Cat B. In contrast to cathepsins, the activities of endogenous cysteine peptidase inhibitors (CPIs), including stefins, cystatins and kininogens, were significantly higher in benign and atypical meningioma cell extracts than in malignant meningioma, and low-grade compared to high-grade astrocytoma. However, very low levels of stefins A and B were found in meningioma and glioblastoma tissues. Further studies on the expression levels and balance between cysteine endopeptidases (CPs) and CPIs would improve the clinical application of cathepsins in prognosis, which would lead to more-informed therapeutic strategies.

Evaluating time series forecasting models: An empirical study on performance estimation methods

Abstract: Performance estimation aims at estimating the loss that a predictive model will incur on unseen data. These procedures are part of the pipeline in every machine learning project and are used for assessing the overall generalisation ability of predictive models. In this paper we address the application of these methods to time series forecasting tasks. For independent and identically distributed data the most common approach is cross-validation. However, the dependency among observations in time series raises some caveats about the most appropriate way to estimate performance in this type of data and currently there is no settled way to do so. We compare different variants of cross-validation and of out-of-sample approaches using two case studies: One with 62 real-world time series and another with three synthetic time series. Results show noticeable differences in the performance estimation methods in the two scenarios. In particular, empirical experiments suggest that cross-validation approaches can be applied to stationary time series. However, in real-world scenarios, when different sources of non-stationary variation are at play, the most accurate estimates are produced by out-of-sample methods that preserve the temporal order of observations.