Supramolecular template-directed synthesis of triazole oligomers.

Sandwich complexes formed by two zinc porphyrins and a diamine ligand (DABCO) have been used as a supramolecular template to direct the synthesis of triazole oligomers. Monomer units equipped with two polymerizable functional groups, an alkyne and an azide, were attached to the template via ester bonds between a phenol unit on the monomer and benzoic acid units on the porphyrin. Self-assembly of the zinc porphyrins by addition of DABCO led to a supramolecular complex containing four of the monomer units, two on each porphyrin. CuAAC oligomerisation was carried out in the presence of a chain capping agent to prevent intermolecular reactions between the templated products, which carry reactive chain ends. The templated-directed oligomerisation resulted in selective formation of a duplex, which contains two identical chains of triazole oligomers connecting the porphyrin linkers. The effective molarity for the intramolecular CuAAC reactions on the template is 3-9 mM, and because the triazole backbone has a direction, the product duplex was obtained as a 4 : 1 mixture of the parallel and antiparallel isomers. Hydrolysis of the ester bonds connecting the oligomers to the template gave a single product, the phenol 2-mer, in excellent yield. The introduction of a supramolecular element into the template considerably broadens the scope of the covalent template-directed oligomerisation methodology that we previously developed for the replication of sequence information in synthetic oligomers.

High-Fidelity Sequence-Selective Duplex Formation by Recognition-Encoded Melamine Oligomers.

Melamine oligomers composed of repeating triazine-piperidine units and equipped with phenol and phosphine oxide side-chains form H-bonded duplexes. The melamine backbone provides sufficient rigidity to prevent intramolecular folding of oligomers up to three recognition units in length, leading to reliable duplex formation between sequence complementary oligomers. NMR spectroscopy and isothermal titration calorimetry (ITC) were used to characterize the self-assembly properties of the oligomers. For length-complementary homo-oligomers, duplex formation in toluene is characterized by an increase in stability of an order of magnitude for every base-pair added to the chain. NMR spectra of dilute solutions of the AD 2-mer show that intramolecular H-bonding between neighboring recognition units on the chain (1,2-folding) does not occur. NMR spectra of dilute solutions of both the AAD and the ADD 3-mer show that 1,3-folding does not take place either. ITC was used to characterize interactions between all pairwise combinations of the six different 3-mer sequences, and the sequence complementary duplexes are approximately an order of magnitude more stable than duplexes with a single base mismatch. High-fidelity duplex formation combined with the synthetic accessibility of the monomer building blocks makes these systems attractive targets for further investigation.

Molecular Transformer: A Model for Uncertainty-Calibrated Chemical Reaction Prediction.

Organic synthesis is one of the key stumbling blocks in medicinal chemistry. A necessary yet unsolved step in planning synthesis is solving the forward problem: Given reactants and reagents, predict the products. Similar to other work, we treat reaction prediction as a machine translation problem between simplified molecular-input line-entry system (SMILES) strings (a text-based representation) of reactants, reagents, and the products. We show that a multihead attention Molecular Transformer model outperforms all algorithms in the literature, achieving a top-1 accuracy above 90% on a common benchmark data set. Molecular Transformer makes predictions by inferring the correlations between the presence and absence of chemical motifs in the reactant, reagent, and product present in the data set. Our model requires no handcrafted rules and accurately predicts subtle chemical transformations. Crucially, our model can accurately estimate its own uncertainty, with an uncertainty score that is 89% accurate in terms of classifying whether a prediction is correct. Furthermore, we show that the model is able to handle inputs without a reactant-reagent split and including stereochemistry, which makes our method universally applicable.

Molecular Transformer unifies reaction prediction and retrosynthesis across pharma chemical space.

Predicting how a complex molecule reacts with different reagents, and how to synthesise complex molecules from simpler starting materials, are fundamental to organic chemistry. We show that an attention-based machine translation model - Molecular Transformer - tackles both reaction prediction and retrosynthesis by learning from the same dataset. Reagents, reactants and products are represented as SMILES text strings. For reaction prediction, the model "translates" the SMILES of reactants and reagents to product SMILES, and the converse for retrosynthesis. Moreover, a model trained on publicly available data is able to make accurate predictions on proprietary molecules extracted from pharma electronic lab notebooks, demonstrating generalisability across chemical space. We expect our versatile framework to be broadly applicable to problems such as reaction condition prediction, reagent prediction and yield prediction.