Increasing Complexity: A Practical Synthetic Approach to Three-Dimensional,Cyclic Sulfoximines and First Insights into Their in Vitro Properties

A short synthetic approach with broad scope to access five- to seven-membered cyclic sulfoximines in only two to three steps from readily available thiophenols is reported. Thus, simple building blocks were converted to complex molecular structures by a sequence of S-alkylation and one-pot sulfoximine formation, followed by intramolecular cyclization. Seventeen structurally diverse cyclic sulfoximines were prepared in high overall yields. In vitro evaluation of these underrepresented, three-dimensional, cyclic sulfoximines with respect to properties relevant to medicinal chemistry did not reveal any intrinsic flaw for application in drug discovery.     

Accurate solubility prediction with error bars for electrolytes: a machine learning approach

Accurate in silico models for predicting aqueous solubility are needed in drug design and discovery and many other areas of chemical research. We present a statistical modeling of aqueous solubility based on measured data, using a Gaussian Process nonlinear regression model (GPsol). We compare our results with those of 14 scientific studies and 6 commercial tools. This shows that the developed model achieves much higher accuracy than available commercial tools for the prediction of solubility of electrolytes. On top of the high accuracy, the proposed machine learning model also provides error bars for each individual prediction.     

Asymmetric AZA-BODIPY with Optical Gain in the Near-Infrared Region

In recent years, there has been a lot of interest in the development of organic compounds emitting in the near-infrared (NIR) region due to their stimulating applications, such as biosensing and light detection and ranging (LiDAR). Moreover, a lot of effort has been devoted to finding organic emitters with optical gain in the NIR region for lasing applications. In this paper, we present the ultrafast spectroscopy of an asymmetric AZA-BODIPY molecule that shows relevant photophysical changes moving from a diluted solution to a concentrated solution and to a spin-coated film. The diluted solution and the spin-coated film show a bleaching band and a stimulated emission band in the visible region, while the very concentrated solution displays a broad (150 nm) and long-living (more than 400 ps) optical gain band in the NIR region, centered at 900 nm. Our results pave the way for a new organic laser system in a near-infrared spectral region.     

Impact of Pore Clogging by Bacteria on Underground Hydrogen Storage

The idea of underground storage of hydrogen exists today in several forms. One of the most promising methods is the option of underground methanation, which consists of injecting H₂ and CO₂ into an underground porous reservoir (aquifer) and converting them into methane by means of methanogenic bacteria that initiate the methanation reaction. However, due to their activities, the high accumulation of bacteria in the pore walls causes pore-clogging (microbial-induced clogging); one of the main problems that can become an obstacle to the implementation of this idea. In this paper, we develop a conceptual model of bio-clogging, which consists of several stages of attachment to pore walls, detachment from the walls and pore plugging by the biomass growth. This model was built into the numerical model of multicomponent two-phase flow. The effect of bio-clogging on gas movement in the storage was analyzed numerically. It has been found that bio-clogging reduces vertical rise of hydrogen and results in more uniform radial gas penetration into the reservoir.

     

Estimating the domain of applicability for machine learning QSAR models: a study on aqueous solubility of drug discovery molecules

We investigate the use of different Machine Learning methods to construct models for aqueous solubility. Models are based on about 4000 compounds, including an in-house set of 632 drug discovery molecules of Bayer Schering Pharma. For each method, we also consider an appropriate method to obtain error bars, in order to estimate the domain of applicability (DOA) for each model. Here, we investigate error bars from a Bayesian model (Gaussian Process (GP)), an ensemble based approach (Random Forest), and approaches based on the Mahalanobis distance to training data (for Support Vector Machine and Ridge Regression models). We evaluate all approaches in terms of their prediction accuracy (in cross-validation, and on an external validation set of 536 molecules) and in how far the individual error bars can faithfully represent the actual prediction error.     

A Unified Experimental/Theoretical Description of the Ultrafast Photophysics of Single and Double Thionated Uracils

Photoinduced processes in thiouracil derivatives have lately attracted considerable attention due to their suitability for innovative biological and pharmacological applications. Here, sub-20 fs broadband transient absorption spectroscopy in the near-UV are combined with CASPT2/MM decay path calculations to unravel the excited-state decay channels of water solvated 2-thio and 2,4-dithiouracil. These molecules feature linear absorption spectra with overlapping ππ* bands, leading to parallel decay routes which we systematically track for the first time. The results reveal that different processes lead to the triplet states population, both directly from the ππ* absorbing state and via the intermediate nπ* dark state. Moreover, the 2,4-dithiouracil decay pathways is shown to be strongly correlated either to those of 2- or 4-thiouracil, depending on the sulfur atom on which the electronic transition localizes.     

Estimating the domain of applicability for machine learning QSAR models: a study on aqueous solubility of drug discovery molecules

We investigate the use of different Machine Learning methods to construct models for aqueous solubility. Models are based on about 4000 compounds, including an in-house set of 632 drug discovery molecules of Bayer Schering Pharma. For each method, we also consider an appropriate method to obtain error bars, in order to estimate the domain of applicability (DOA) for each model. Here, we investigate error bars from a Bayesian model (Gaussian Process (GP)), an ensemble based approach (Random Forest), and approaches based on the Mahalanobis distance to training data (for Support Vector Machine and Ridge Regression models). We evaluate all approaches in terms of their prediction accuracy (in cross-validation, and on an external validation set of 536 molecules) and in how far the individual error bars can faithfully represent the actual prediction error.