Blinded Predictions and Post Hoc Analysis of the Second Solubility Challenge Data: Exploring Training Data and Feature Set Selection for Machine and Deep Learning Models.

Accurate methods to predict solubility from molecular structure are highly sought after in the chemical sciences. To assess the state of the art, the American Chemical Society organized a "Second Solubility Challenge" in 2019, in which competitors were invited to submit blinded predictions of the solubilities of 132 drug-like molecules. In the first part of this article, we describe the development of two models that were submitted to the Blind Challenge in 2019 but which have not previously been reported. These models were based on computationally inexpensive molecular descriptors and traditional machine learning algorithms and were trained on a relatively small data set of 300 molecules. In the second part of the article, to test the hypothesis that predictions would improve with more advanced algorithms and higher volumes of training data, we compare these original predictions with those made after the deadline using deep learning models trained on larger solubility data sets consisting of 2999 and 5697 molecules. The results show that there are several algorithms that are able to obtain near state-of-the-art performance on the solubility challenge data sets, with the best model, a graph convolutional neural network, resulting in an RMSE of 0.86 log units. Critical analysis of the models reveals systematic differences between the performance of models using certain feature sets and training data sets. The results suggest that careful selection of high quality training data from relevant regions of chemical space is critical for prediction accuracy but that other methodological issues remain problematic for machine learning solubility models, such as the difficulty in modeling complex chemical spaces from sparse training data sets.

Flip-flop asymmetry of cholesterol in model membranes induced by thermal gradients.

Lipid asymmetry is a crucial property of biological membranes and significantly influences their physical and mechanical properties. It is responsible for maintaining different chemical environments on the external and internal surfaces of cells and organelles and plays a vital role in many biological processes such as cell signalling and budding. In this work we show, using non-equilibrium molecular dynamics (NEMD) simulations, that thermal fields can induce lipid asymmetry in biological membranes. We focus our investigation on cholesterol, an abundant lipid in the plasma membrane, with a rapid flip-flop rate, significantly influencing membrane properties. We demonstrate that thermal fields induce membrane asymmetry with cholesterol showing thermophobic behaviour and therefore accumulating on the cold side of the membrane. This work highlights a possible experimental route to preparing and controlling asymmetry in synthetic membranes.

Unilateral phacoemulsification in a captive African elephant (Loxodonta africana).

Background: The following case reports describe the clinical presentation, surgical protocol, post-operative care, and long-term follow-up of an African elephant (Loxodonta Africana) presenting with a unilateral cataract. Case description: A 42-year-old female African elephant presented for the assessment of ocular discomfort and visual deterioration in the left eye. Pre-surgical treatment included topical anti-inflammatory medication for 20 days prior to surgery. On the day of surgery, following anesthetic induction, a two-handed phacoemulsification technique was performed in the left eye. She was left aphakic post-operatively. Nine days post-operatively, the patient had an intact menace response, dazzle reflex, and direct pupillary light reflex. Fundoscopy at that stage was unremarkable. Follow-up information was available for 5 years, from the time of surgery to the present day. Conclusion: Despite remaining aphakic, this case presents a successful visual outcome. To the best of the authors' knowledge, there is no other published report of phacoemulsification in a captive elephant.

Prevalence of antibody seroconversion to Toxoplasma gondii in uveitis and non-uveitis dogs.

OBJECTIVES: To evaluate the prevalence of seroconversion to Toxoplasma gondii in dogs with uveitis and dogs without uveitis. METHODS: In total, 135 dogs were evaluated: 51 dogs were diagnosed with uveitis, and 84 dogs were without uveitis. Latex agglutination tests were performed on all sera, and the results were evaluated. RESULTS: Overall, 7.8 and 6.0 per cent of sera were positive for the presence of anti-T gondii antibodies (dilution ≥1:64) in the groups with uveitis and non-uveitis dogs, respectively. The frequency distribution of variables (positive and negative results in the uveitis and the non-uveitis group of dogs) was tested with Fisher's exact test. There was no statistically significant difference between groups (P=0.73). CLINICAL SIGNIFICANCE: These findings suggest that evidence of exposure to T gondii was not significantly different between uveitis and non-uveitis groups of dogs and that the possible association between exposure to T gondii and canine uveitis requires further investigation. This study is the first to report the seroprevalence of anti-T gondii antibodies in dogs in the UK population and the first to report the seroprevalence of anti-T gondii antibodies in dogs with uveitis.

Structural organization of sterol molecules in DPPC bilayers: a coarse-grained molecular dynamics investigation.

We investigate the structural organization of cholesterol (CHOL) analogues in 1,2-dipalmitoyl-sn-glycero-3-phosphocholine (DPPC) bilayers using coarse-grained molecular dynamics simulations and the MARTINI forcefield. Different sterol molecules are modelled by increasing (CHOLL) or decreasing (CHOLS) the diameter of the sterol beads employed in the MARTINI model of CHOL. At high sterol concentrations, (xsterol = 0.5), typical of liquid ordered phases, we find that the sterol arrangement and sterol-DPPC interactions strongly depend on the sterol size. Smaller sterols (CHOLS and CHOL) form linear clusters, while the larger sterols (CHOLL) arrange themselves into disc shaped clusters. By combining structural and dynamical properties we also investigate the So→ Ld transition for the CHOLL and CHOLS sterols. We show that small changes in the sterol size significantly affect the stability of the gel phase with the gel phase stabilized by the small sterols, but destabilized by large sterols. The general dependence of the phase behaviour of the membrane with sterol content is reminiscent of the one observed in naturally occurring membranes. The relevance of our results to understand current cholesterol-bilayer structural models is discussed.

Engineering folding dynamics from two-state to downhill: application to λ-repressor.

One strategy for reaching the downhill folding regime, primarily exploited for the λ(6-85) protein fragment, consists of cumulatively introducing mutations that speed up folding. This is an experimentally demanding process where chemical intuition usually serves as a guide for the choice of amino acid residues to mutate. Such an approach can be aided by computational methods that screen for protein engineering hot spots. Here we present one such method that involves sampling the energy landscape of the pseudo-wild-type protein and investigating the effect of point mutations on this landscape. Using a novel metric for the cooperativity, we identify those residues leading to the least cooperative folding. The folding dynamics of the selected mutants are then directly characterized and the differences in the kinetics are analyzed within a Markov-state model framework. Although the method is general, here we present results for a coarse-grained topology-based simulation model of λ-repressor, whose barrier is reduced from an initial value of ∼4 k(B)T at the midpoint to ∼1 k(B)T, thereby reaching the downhill folding regime.