Symbolic, neural, and Bayesian machine learning models for predicting carcinogenicity of chemical compounds

Experimental programs have been underway for several years to determine the environmental effects of chemical compounds, mixtures, and the like. Among these programs is the National Toxicology Program (NTP) on rodent carcinogenicity. Because these experiments are costly and time-consuming, the rate at which test articles (i.e., chemicals) can be tested is limited. The ability to predict the outcome of the analysis at various points in the process would facilitate informed decisions about the allocation of testing resources. To assist human experts in organizing an empirical testing regime, and to try to shed light on mechanisms of toxicity, we constructed toxicity models using various machine learning and data mining methods, both existing and those of our own devising. These models took the form of decision trees, rule sets, neural networks, rules extracted from trained neural networks, and Bayesian classifiers. As a training set, we used recent results from rodent carcinogenicity bioassays conducted by the NTP on 226 test articles. We performed 10-way cross-validation on each of our models to approximate their expected error rates on unseen data. The data set consists of physical-chemical parameters of test articles, alerting chemical substructures, salmonella mutagenicity assay results, subchronic histopathology data, and information on route, strain, and sex/species for 744 individual experiments. These results contribute to the ongoing process of evaluating and interpreting the data collected from chemical toxicity studies.     

Modeling NMR chemical shift: A survey of density functional theory approaches for calculating tensor properties

The NMR chemical shift, a six-parameter tensor property, is highly sensitive to the position of the atoms in a molecule. To extract structural parameters from chemical shifts, one must rely on theoretical models. Therefore, a high quality group of shift tensors that serve as benchmarks to test the validity of these models is warranted and necessary to highlight existing computational limitations. Here, a set of 102 13C chemical-shift tensors measured in single crystals, from a series of aromatic and saccharide molecules for which neutron diffraction data are available, is used to survey models based on the density functional (DFT) and Hartree-Fock (HF) theories. The quality of the models is assessed by their least-squares linear regression parameters. It is observed that in general DFT outperforms restricted HF theory. For instance, Becke's three-parameter exchange method and mpw1pw91 generally provide the best predicted shieldings for this group of tensors. However, this performance is not universal, as none of the DFT functionals can predict the saccharide tensors better than HF theory. Both the orientations of the principal axis system and the magnitude of the shielding were compared using the chemical-shift distance to evaluate the quality of the calculated individual tensor components in units of ppm. Systematic shortcomings in the prediction of the principal components were observed, but the theory predicts the corresponding isotropic value more accurately. This is because these systematic errors cancel, thereby indicating that the theoretical assessment of shielding predictions based on the isotropic shift should be avoided.     

Computational approaches for predicting mutant protein stability

Mutations in the protein affect not only the structure of protein, but also its function and stability. Prediction of mutant protein stability with accuracy is desired for uncovering the molecular aspects of diseases and design of novel proteins. Many advanced computational approaches have been developed over the years, to predict the stability and function of a mutated protein. These approaches based on structure, sequence features and combined features (both structure and sequence features) provide reasonably accurate estimation of the impact of amino acid substitution on stability and function of protein. Recently, consensus tools have been developed by incorporating many tools together, which provide single window results for comparison purpose. In this review, a useful guide for the selection of tools that can be employed in predicting mutated proteins’ stability and disease causing capability is provided.     

Cellular biosensing: chemical and genetic approaches

Biosensors have been developed to determine the concentration of specific compounds in situ. They are already widely employed as a practical technology in the clinical and healthcare fields. Recently, another concept of biosensing has been receiving attention: biosensing for the evaluation of molecular potency. The development of this novel concept has been supported by the development of related technologies, as such as molecular design, molecular biology (genetic engineering) and cellular/tissular engineering. This review is addresses this new concept of biosensing and its application to the evaluation of the potency of chemicals in biological systems, in the field of cellular/tissular engineering. Cellular biosensing may provide information on both pharmaceutical and chemical safety, and on drug efficacy in vitro as a screening tool.     

人工智能化学：变革研究范式,加速物质发现

化学科学领域的复杂性和海量数据为人工智能应用提供了契机。人工智能、机器学习、深度学习从海量数据中识别新的化合物,建立新的模型,提出新的理论,正在改变化学物质的发现、转化和功能研究范式,促进重大问题的解决。本文综述了近年来国际上人工智能在化学研究中的重要进展,分析了人工智能化学的主要发展态势。人工智能通过助力化学海量数据挖掘、实现化学实验室智能化和自动化、增强计算化学解决实际问题的能力,推动化学跨越式发展。     

Computational approaches for understanding and predicting the self-assembled peptide hydrogels

Self-assembled peptide hydrogel is a promising biomaterial and has been widely applied in many fields. As a typical self-assembly material, peptide hydrogel exhibits properties different from traditional polymer hydrogel, and has unique features in molecular design, structural elements of hydrogel, and control strategies. With the desire to apply the principles of self-assembly to the design and prediction of peptide hydrogels, there has more and more emphasis on understanding the driving forces and microscopic behaviors involved in the self-assembly process. Computational methods have played an increasingly important role in recent research in helping to reveal the relationship between molecular chemical structure and self-assembly processes as well as assembled morphologies, thus determining the ability of supramolecular gelation. This review aims to summarize the application of computational tools to obtain a better fundamental understanding of the multi-scale structural details of self-assembled peptide hydrogels and to predict the gelation behavior of supramolecular nanofibers. It is expected that researchers will consider using these computational tools when investigating and designing novel peptide hydrogel materials.     

SMIREP: predicting chemical activity from SMILES

Most approaches to structure-activity-relationship (SAR) prediction proceed in two steps. In the first step, a typically large set of fingerprints, or fragments of interest, is constructed (either by hand or by some recent data mining techniques). In the second step, machine learning techniques are applied to obtain a predictive model. The result is often not only a highly accurate but also hard to interpret model. In this paper, we demonstrate the capabilities of a novel SAR algorithm, SMIREP, which tightly integrates the fragment and model generation steps and which yields simple models in the form of a small set of IF-THEN rules. These rules contain SMILES fragments, which are easy to understand to the computational chemist. SMIREP combines ideas from the well-known IREP rule learner with a novel fragmentation algorithm for SMILES strings. SMIREP has been evaluated on three problems: the prediction of binding activities for the estrogen receptor (Environmental Protection Agency's (EPA's) Distributed Structure-Searchable Toxicity (DSSTox) National Center for Toxicological Research estrogen receptor (NCTRER) Database), the prediction of mutagenicity using the carcinogenic potency database (CPDB), and the prediction of biodegradability on a subset of the Environmental Fate Database (EFDB). In these applications, SMIREP has the advantage of producing easily interpretable rules while having predictive accuracies that are comparable to those of alternative state-of-the-art techniques.     

Marine derived polysaccharides for biomedical applications: chemical modification approaches

Polysaccharide-based biomaterials are an emerging class in several biomedical fields such as tissue regeneration, particularly for cartilage, drug delivery devices and gelentrapment systems for the immobilization of cells. Important properties of the polysaccharides include controllable biological activity, biodegradability, and their ability to form hydrogels. Most of the polysaccharides used derive from natural sources; particularly, alginate and chitin, two polysaccharides which have an extensive history of use in medicine, pharmacy and basic sciences, and can be easily extracted from marine plants (algae kelp) and crab shells, respectively. The recent rediscovery of poly-saccharidebased materials is also attributable to new synthetic routes for their chemical modification, with the aim of promoting new biological activities and/or to modify the final properties of the biomaterials for specific purposes. These synthetic strategies also involve the combination of polysaccharides with other polymers. A review of the more recent research in the field of chemical modification of alginate, chitin and its derivative chitosan is presented. Moreover, we report as case studies the results of our recent work concerning various different approaches and applications of polysaccharide-based biomaterials, such as the realization of novel composites based on calcium sulphate blended with alginate and with a chemically modified chitosan, the synthesis of novel alginate-poly(ethylene glycol) copolymers and the development of a family of materials based on alginate and acrylic polymers of potential interest as drug delivery systems.     

Advances and approaches for chemical recycling of plastic waste

The global production and consumption of plastics has increased at an alarming rate over the last few decades. The accumulation of pervasive and persistent waste plastic has concomitantly increased in landfills and the environment. The societal, ecological, and economic problems of plastic waste/pollution demand immediate and decisive action. In 2015, only 9% of plastic waste was successfully recycled in the United States. The major current recycling processes focus on the mechanical recycling of plastic waste; however, even this process is limited by the sorting/pretreatment of plastic waste and degradation of plastics during the process. An alternative to mechanical processes is chemical recycling of plastic waste. Efficient chemical recycling would allow for the production of feedstocks for various uses including fuels and chemical feedstocks to replace petrochemicals. This review focuses on the most recent advances for the chemical recycling of three major polymers found in plastic waste: PET, PE, and PP. Commercial processes for recycling hydrolysable polymers like polyesters or polyamides, polyolefins, or mixed waste streams are also discussed.     

Symmetry rules for predicting the course of chemical reactions

Rules are given for a reaction of any molecularity which show whether it is allowed or forbidden by orbital symmetry properties. The rules are based on a rigorous treatment, but correspond to simple concepts in practice. A rough MO sequence for the reactants is all that is needed.

---

Zusammenfassung Für Reaktionen beliebiger Molekularität werden Regeln dafür angegeben, ob die Reaktion aus Symmetriegründen erlaubt oder verboten ist. Diese Regeln werden streng hergeleitet, sind jedoch in der Praxis einfach anwendbar. Benötigt wird lediglich die Reihenfolge der MO der Reaktionen.

---

Résumé On donne des règles de sélection pour une réaction de molécularité arbitraire selon les propriétés de symétrie orbitale. Les règles sont basées sur un traitement rigoureux mais correspondent en practique à des concepts simples. On a seulement besoin de connaître l'ordre des orbitales moléculaires des réactants.

     

High pressure in solid state chemistry: Combined experimental and modeling approaches for assessing and predicting properties

The thermodynamic pressure parameter has been thoroughly used with mastership by Gérard Demazeau throughout his rich career in solid state chemistry and materials sciences and more recently in biosciences. After a review of such works, focus is made in this topical article on his contribution together with his team in the field of hard materials based on light elements B, C, N with a proposition of a new ultra-hard carbon nitride C₂N on one hand and on the structural transformations under high pressures of perovskite into postperovskite with a change of dimensionality from 3D to 2D and related oxides, regarding the arrangement of octahedra, on the other hand. Investigation and concepts first arising from experimental observables are shown to be aided and accelerated via first principles calculations of energy and energy-related quantities.     

Theoretical approaches for predicting the color of rigid dyes in solution

Aiming at developing an affordable and easily implementable computational protocol for routine prediction of spectral properties of rigid molecular dyes, density functional theory, and time‐dependent density functional theory were used in conjunction with a vibronic coupling scheme for band shape estimate. To predict the perceived color of molecules in solution, a model has been setup linking the UV‐vis spectra predicted at ab initio level to the L *a *b * colorimetric parameters. The results show that a mixed protocol, implying the use of a global hybrid functional for the prediction of adiabatic energy differences and a range separated hybrid for the prediction of potential energy curvature, allows perceived colors to be quantitatively predicted, as demonstrated by the comparison of L *a *b * colorimetric parameters obtained from computed and experimental spectra. © 2017 Wiley Periodicals, Inc.     

FHA domain-ligand interactions: importance of integrating chemical and biological approaches

Combinatorial library screens based on binding affinity may preferentially select ligands with ability for ionic interactions and miss the biologically relevant ligands that bind more weakly with predominantly hydrophobic interactions.     

SAR by ILOEs: an NMR-based approach to reverse chemical genetics

Reverse chemical genetics is an emerging technique that makes use of small molecule inhibitors to characterize how a protein functions. In this regard, we have developed an NMR-based approach (SAR by ILOEs) that enables the identification of high affinity ligands for a given protein target without the need of a specific assay. Our approach is of general applicability and could result very powerful in reverse chemical-genetics studies, target validation, and lead discovery. We report a recent application on the design and synthesis of compounds that inhibit protein-membrane interactions.     

Optical multiple chemical sensing: status and current challenges

Multiple optical sensors for chemical species are sensitive, non-toxic and non-invasive and enable spatially and temporally resolved multianalyte detection. Recent advances are highlighted with a focus on fluorescence-based methods and the biologically and clinically important analytes oxygen, pH, carbon dioxide and temperature. Indicator chemistries such as permeation-selective microbeads and nanoparticles allow the production of microscopically homogeneous sensor layers. The use of combinations of spectral discrimations along with time-resolved monitoring schemes based on luminescence lifetime or intensity-lifetime ratios enables all-optical real-time multianalyte determination.     

New enzymatic and chemical approaches to enantiopure etodolac

(+)- and (−)-etodolac enantiomers were prepared both by classical resolution via crystallisation of diastereoisomeric salts with (+) and (−)--methylbenzylamine, and by suitable manipulation of derivatives (−)-3- and (+)-4, obtained by lipase-catalysed kinetic resolution of racemic 3 X-ray diffraction analysis of the 4-bromobenzoate derivative of (+)-3, obtained from enantiopure acetate (+)-4, allowed us to determine the absolute (R) configuration of (−)-etodolac.