Accurate prediction of complex formulation viscosity is crucial for the lubricant industry and engineering processes. However, developing generalizable models remains challenging due to the complexity of industrial mixtures and the scarcity of public experimental data, often resulting from corporate proprietary restrictions. This study presents a machine-learning workflow for viscosity estimation from a proprietary pseudonymized dataset, showing that robust models can be trained under data-protection limitations. We evaluated multiple machine learning architectures, including Decision Tree, Random Forest, Support Vector Regression, and Gradient Boosting, employing feature selection strategies adapted to pseudonymized data. Additionally, the Walther formula was used as a physics-based reference. While this formula required deanonymized multi-temperature viscosity data, it proved valuable for identifying experimental outliers. Among the evaluated models, Gradient Boosting demonstrated superior performance, achieving a median absolute percentage error of 16.2% on the held-out test set. These results indicate that machine learning can support viscosity screening for complex industrial fluids without requiring full compositional transparency.
Rational modification and enhancement of protein function are extremely important both for understanding the fundamental principles of operation of natural macromolecules and for practical applications. Recent flourishing of machine learning-based methods in structural biology led to noteworthy advances in protein structure prediction tasks, which in turn enabled development of versatile protein engineering approaches. Among them, the methods aimed at predicting the sequence that would fold into a conditioned structure proved particularly useful. Here, we review the diversity of these methods, their applications in basic and applied research, and discuss the challenges that remain to be overcome in the future.
We say that a subset M of R-n is exponentially Ramsey if there exists epsilon > 0 and n(0) such that chi(R-n, M) > (1 + epsilon)(n) for any n > n(0), where chi(R-n, M) stands for the minimum number of colors in a coloring of Rnsuch that no copy of M is monochromatic. One important result in Euclidean Ramsey theory is due to Frankl and Rodl, and states the following (under some mild extra conditions): if both N-1 and N-2 are exponentially Ramsey then so is their Cartesian product. Applied several times to simple two-point sets N-i, this result implies that any subset M of a 'hyperrectangle' N(1)x ... x N-k is exponentially Ramsey. However, generally, such 'embeddings' of M result in very inefficient bounds on the aforementioned epsilon. In this paper, we present another way of combining exponentially Ramsey sets, which gives much better estimates in some important cases. In particular, we show that the chromatic number of Rnwith a forbidden equilateral triangle satisfies chi(R-n, Delta) >= (1.0742...+ o(1))(n), greatly improving upon the previous constant 1.0144. We also obtain similar strong results for regular simplices of larger dimensions, as well as for related geometric Ramsey-type questions in Manhattan norm. We then show that the same technique implies several interesting corollaries in other combinatorial problems. In particular, we give an explicit upper bound on the size of a family F subset of 2([n]) that contains no weak k-sunflowers, i.e. no collection of k sets with pairwise intersections of the same size. This bound improves upon previously known results for all k >= 4. Finally, we also present a simple deduction of the (other) celebrated Frankl-Rodl theorem from an earlier result of Frankl and Wilson. It gives probably the shortest known proof of Frankl and Rodl result with the most efficient bounds.
The conserved eukaryotic translation initiation factor 5A (eIF5A), which contains hypusine as a post-translational modification, plays a critical role in eukaryotic cell physiology. For instance, the deletion of the eif5A gene in Saccharomyces cerevisiae is lethal to cells. Besides, the inhibition of hypusination enzymes severely suppresses the ability of the cells to divide and halts the growth of mammalian cell populations. Therefore, the study of the structural features of eIF5A and its complexes with the enzymes deoxyhypusine synthase (DHS) and deoxyhypusine hydroxylase (DOHH) from the pathogenic yeast-like fungi Candida albicans, which catalyze the post-translational modification of eIF5A, will help in searching for new targets for the development of new antimycotics. This study presents an optimized protocol for the isolation and purification of the eIF5A protein from C. albicans and the structural analysis of this protein and its complex with the DHS enzyme by small-angle X-ray scattering.
We calculate the energies and investigate the physical properties of the neutral vector K*0 mesons in external strong abelian magnetic field. We explore how the magnetic moment of the neutral vector K* mesons depends on the ratio of the bare strange and light quark masses ms/md. The extrapolation of g-factor to the physical pion mass was performed at ms/md = 20. We also calculated the magnetic dipole polarizability of K*0-meson for the sz = −1 case.
