Rafael Gómez-Bombarelli

Automatic Chemical Design Using a Data-Driven Continuous Representation of Molecules

We report a method to convert discrete representations of molecules to and from a multidimensional continuous representation. This model allows us to generate new molecules for efficient exploration and optimization through open-ended spaces of chemical compounds. A deep neural network was trained on hundreds of thousands of existing chemical structures to construct three coupled functions: an encoder, a decoder, and a predictor. The encoder converts the discrete representation of a molecule into a real-valued continuous vector, and the decoder converts these continuous vectors back to discrete molecular representations. The predictor estimates chemical properties from the latent continuous vector representation of the molecule. Continuous representations of molecules allow us to automatically generate novel chemical structures by performing simple operations in the latent space, such as decoding random vectors, perturbing known chemical structures, or interpolating between molecules. Continuous representations also allow the use of powerful gradient-based optimization to efficiently guide the search for optimized functional compounds. We demonstrate our method in the domain of drug-like molecules and also in a set of molecules with fewer that nine heavy atoms.

Computational Calculation of Equilibrium Constants: Addition to Carbonyl Compounds

Hydration reactions are relevant for understanding many organic mechanisms. Since the experimental determination of hydration and hemiacetalization equilibrium constants is fairly complex, computational calculations now offer a useful alternative to experimental measurements. In this work, carbonyl hydration and hemiacetalization constants were calculated from the free energy differences between compounds in solution, using absolute and relative approaches. The following conclusions can be drawn: (i) The use of a relative approach in the calculation of hydration and hemiacetalization constants allows compensation of systematic errors in the solvation energies. (ii) On average, the methodology proposed here can predict hydration constants within +/- 0.5 log K(hyd) units for aldehydes. (iii) Hydration constants can be calculated for ketones and carboxylic acid derivatives within less than +/- 1.0 log K(hyd), on average, at the CBS-Q level of theory. (iv) The proposed methodology can predict hemiacetal formation constants accurately at the MP2 6-31++G(d,p) level using a common reference. If group references are used, the results obtained using the much cheaper DFT-B3LYP 6-31++G(d,p) level are almost as accurate. (v) In general, the best results are obtained if a common reference for all compounds is used. The use of group references improves the results at the lower levels of theory, but at higher levels, this becomes unnecessary.

Computational Study of the Acid Dissociation of Esters and Lactones. A Case Study of Diketene

A computational study of the aqueous pK(a) of some saturated and unsaturated cyclic and linear esters and ketones was carried out at the DFT-B3LYP 6-31++G(2df,2pd), CBS-Q, and G2 levels, with the integral equation formalism polarizable continuum model for solvation, using a proton exchange mechanism. The influence of unsaturation, position of the double bond, and cyclization were studied. The computational results show that (a) in all cases studied except that of diketene (4-methylene-2-oxetanone), the alpha-beta unsaturated isomer is 20-30 kJ mol(-1) lower in energy that the beta-gamma unsaturated one; (b) alpha-beta unsaturation lowers the pK(a) of an ester approximately 6 units, whereas beta-gamma unsaturation lowers it by approximately 10 units, and cyclization lowers the pK(a) by approximately 3 units. In order to check the predictive power of the methodology, the acid dissociation constant of diketene in water was measured via kinetic study of its base-catalyzed hydrolysis. The pK(a) value obtained (15.2 +/- 0.3) is in keeping with the expected value for a beta-gamma unsaturated beta-lactone. This low value also suggests that deprotonated diketene does not interconvert to a more stable, less acidic alpha-beta unsaturated isomer, which is also consistent with computational results.

Mechanisms of Lactone Hydrolysis in Neutral and Alkaline Conditions

The neutral and base-catalyzed hydrolysis of nine carboxylic acid esters was studied using a hybrid supermolecule-PCM approach including six explicit water molecules. The molecules studied included two linear esters, four β-lactones, two γ-lactones, and one δ-lactone: ethyl acetate and methyl formate, β-propiolactone, β-butyrolactone, β-isovalerolactone, diketene (4-methyleneoxetan-2-one), γ-butyrolactone, 2(5H)-furanone, and δ-valerolactone. DFT and ab initio methods were used to analyze the features of the various possible hydrolysis mechanisms. For all compounds, reasonable to very good qualitative and quantitative agreement with experimental work was found, and evidence is provided to support long-standing hypotheses regarding the role of solvent molecule as a base catalyst. In addition, novel evidence is presented for the existence of an elimination-addition mechanism in the basic hydrolysis of diketene. A parallel work addresses the acid-catalyzed hydrolysis of lactones.

Potential of the NBP Method for the Study of Alkylation Mechanisms: NBP as a DNA-Model

Alkylating agents are considered to be archetypal carcinogens. One suitable technique to evaluate the activity of alkylating compounds is the NBP assay. This method is based on the formation of a chromophore in the reaction between the alkylating agent and the nucleophile 4-(p-nitrobenzyl)pyridine (NBP), a trap for alkylating agents with nucleophilic characteristics similar to those of DNA bases. NBP is known to react with strong and weak alkylating agents, and much insight into such alkylation mechanisms in vivo can be gained from kinetic study of some alkylation reactions in vitro. Since 1925, the NBP assay has evolved from being a qualitative, analytical tool to becoming a useful physicochemical method that not only allows the rules of chemical reactivity that govern electrophilicity and nucleophilicity to be applied to the reaction of DNA with alkylating agents but also helps to understand some significant relationships between the structure of many alkylation substrates (including DNA) and their chemical and biological responses. Given that advances in this area have the potential to yield both fundamental and practical advances in chemistry, biology, predictive toxicology, and anticancer drug development, this review is designed to provide an overview of the evolution of the NBP method from its early inception until its recent kinetic-mechanistic approach, which allows the pros and cons of NBP as a DNA-model to be analyzed. The validity of NBP as a nucleophilicity model for DNA in general and the position of guanosine at N7 in particular are discussed.

An Alternative Host Material for Long‐Lifespan Blue Organic Light‐Emitting Diodes Using Thermally Activated Delayed Fluorescence

It has been challenging to find stable blue organic light emitting diodes (OLEDs) that rely on thermally activated delayed fluorescence (TADF). Lack of stable host materials well‐fitted to the TADF emitters is one of the critical reasons. The most popular host for blue TADF, bis[2‐(diphenylphosphino)phenyl] ether oxide (DPEPO), leads to unrealistically high maximum external quantum efficiency. DPEPO is however an unstable material and has a poor charge transporting ability, which in turn induces an intrinsic short OLED operating lifespan. Here, an alternative host material is introduced which educes the potential efficiency and device lifespan of given TADF emitters with the appropriateness of replacing the most popular host material, DPEPO, in developing blue TADF emitters. It simultaneously provides much longer device lifespan and higher external quantum efficiency at a practical brightness due to its high material stability and electron‐transport‐type character well‐fitted for hole‐transport‐type TADF emitters.

DNA-damaging disinfection byproducts: alkylation mechanism of mutagenic mucohalic acids.

Hydroxyhalofuranones form a group of genotoxic disinfection byproduct (DBP) of increasing interest. Among them, mucohalic acids (3,4-dihalo-5-hydroxyfuran-2(5H)-one, MXA) are known mutagens that react with nucleotides, affording etheno, oxaloetheno, and halopropenal derivatives. Mucohalic acids have also found use in organic synthesis due to their high functionalization. In this work, the alkylation kinetics of mucochloric and mucobromic acids with model nucleophiles aniline and NBP has been studied experimentally. Also, the alkylation mechanism of nucleosides by MXA has been studied in silico. The results described allow us to reach the following conclusions: (i) based on the kinetic and computational evidence obtained, a reaction mechanism was proposed, in which MXA react directly with amino groups in nucleotides, preferentially attacking the exocyclic amino groups over the endocyclic aromatic nitrogen atoms; (ii) the suggested mechanism is in agreement with both the product distribution observed experimentally and the mutational pattern of MXA; (iii) the limiting step in the alkylation reaction is addition to the carbonyl group, subsequent steps occurring rapidly; and (iv) mucoxyhalic acids, the hydrolysis products of MXA, play no role in the alkylation reaction by MXA.

Sorbate—Nitrite Interactions: Acetonitrile Oxide as an Alkylating Agent

Because chemical species with DNA-damaging and mutagenic activity are formed in sorbate-nitrite mixtures and because sorbic acid sometimes coexists with nitrite occurring naturally or incorporated as a food additive, the study of sorbate-nitrite interactions is important. Here, the alkylating potential of the products resulting from such interactions was investigated. Drawn were the following conclusions: (i) Acetonitrile oxide (ACNO) is the compound responsible for the alkylating capacity of sorbate-nitrite mixtures; (ii) ACNO alkylates 4-(p-nitrobenzyl)pyridine (NBP), a trap for alkylating agents with nucleophilic characteristics similar to those of DNA bases, forming an adduct (AD; epsilon = 1.4 x 10(4) M(-1) cm(-1); lambda = 519 nm); (iii) the NBP alkylation reaction complies with the rate equation, r = d[AD]/dt = k(alk)(ACNO)[ACNO][NBP]-k(hyd)(AD)[AD], k(alk)(ACNO) being the NBP alkylation rate constant for ACNO and k(hyd)(AD) the rate constant for the adduct hydrolysis reaction; (iv) the small fraction of ACNO forming the adduct with NBP, as well as the small magnitude of the quotient (k(alk) (ACNO)/k(hyd)(ACNO)) as compared with those reported for other alkylating agents, such as some lactones and N-alkyl-N-nitrosoureas, reveals the ACNO effective alkylating capacity to be less significant; (v) the low value of the NBP-ACNO adduct life (defined as the total amount of adduct present along the progression of the NBP alkylation per unit of alkylating agent concentration) points to the high instability of this adduct; and (vi) the obtained results are in accordance with the low carcinogenicity of ACNO.

Vibration-assisted resonance in photosynthetic excitation-energy transfer

Understanding how the effectiveness of natural photosynthetic energy-harvesting systems arises from the interplay between quantum coherence and environmental noise represents a significant challenge for quantum theory. Recently it has begun to be appreciated that discrete molecular vibrational modes may play an important role in the dynamics of such systems. Here we present a microscopic mechanism by which intramolecular vibrations may be able to contribute to the efficiency and directionality of energy transfer. Excited vibrational states create resonant pathways through the system, supporting fast and efficient energy transport. Vibrational damping together with the natural downhill arrangement of molecular energy levels gives intrinsic directionality to the energy flow. Analytical and numerical results demonstrate a significant enhancement of the efficiency and directionality of energy transport that can be directly related to the existence of resonances between vibrational and excitonic levels.

Biocatalytic oxidation of phenolic compounds by bovine methemoglobin in the presence of H2O2: quantitative structure-activity relationships.

In the present work, 13 p-substituted phenols with different functional groups have been systematically evaluated as metHb substrates by means of HPLC analysis. Non-hyperbolic kinetics were observed and Hill coefficients in the 0.37-1.00 range were obtained. The catalytic constants and the Hill coefficients were found to be quantitatively correlated with two independent variables: the energy level of the highest-occupied molecular orbital (E(HOMO)), which describes the intrinsic redox activity of the substrates and the pK(a)-values, which are related to substrate ionization. Oxygen evolution in the presence of each phenol derivative was also measured, and good correlation between peroxidase-like and catalase-like activities of the protein was observed. It is also shown that bovine metHb, although less active than other peroxidases, may represent a good alternative from an economical point of view for phenol removal processes. The equations here obtained may serve as a basis to further explore the potential use of metHb-mediated reactions in the treatment of phenols in wastewaters and to predict which phenol will be removed most efficiently under this treatment with satisfactory reliability.