Evaluating and clustering retrosynthesis pathways with learned strategy

With recent advances in the computer-aided synthesis planning (CASP) powered by data science and machine learning, modern CASP programs can rapidly identify thousands of potential pathways for a given target molecule. However, the lack of a holistic pathway evaluation mechanism makes it challenging to systematically prioritize strategic pathways except for using some simple heuristics. Herein, we introduce a data-driven approach to evaluate the relative strategic levels of retrosynthesis pathways using a dynamic tree-structured long short-term memory (tree-LSTM) model. We first curated a retrosynthesis pathway database, containing 238k patent-extracted pathways along with ∼55 M artificial pathways generated from an open-source CASP program, ASKCOS. The tree-LSTM model was trained to differentiate patent-extracted and artificial pathways with the same target molecule in order to learn the strategic relationship among single-step reactions within the patent-extracted pathways. The model achieved a top-1 ranking accuracy of 79.1% to recognize patent-extracted pathways. In addition, the trained tree-LSTM model learned to encode pathway-level information into a representative latent vector, which can facilitate clustering similar pathways to help illustrate strategically diverse pathways generated from CASP programs.

Tree-structured long short-term memory neural model learns to understand the retrosynthesis design strategies from patent-extracted retrosynthetic pathway data.     

The ligand, the metal and the 'Holey'-host: Synthesis, structural and magnetic characterisation of Co(II), Ni(II) and Mn(II) metal-organic frameworks incorporating 4,4'-dicarboxy-2,2'-bipyridine

We report herein the single crystal structures of four metal-organic framework complexes incorporating the 4,4'-dicarboxy-2,2'-bipyridine ligand, H(2)dcbp: alpha-[Co(dcbp)(H(2)O)(2)], 1; beta-[Co(dcbp)(H(2)O)(2)], 2, [Ni(dcbp)(H(2)O)(2)], and [[Mn(dcbp)].1/2DEF], 4 (DEF = diethylformamide). In each complex the ligand is deprotonated giving neutral species with 1:1 stoichiometry that form three-dimensional coordination polymers. Supramolecular isomerism (polymorphism) in 1 and 2 arises from the different ligand connectivity around the octahedral Co(II) centres. The two coordinated water molecules in 1 occupy cis positions, which are trans to the chelating bipyridine nitrogen atoms, leaving the carboxylate oxygen atoms in axial trans positions. In 2 all like donors occupy cis positions. Different modes of carboxylate coordination in 1 and 2 give dissimilar network topologies. A rare example of two interpenetrating 6(4)8(2)-b (quartz-like) chiral networks in 1 results from both dcbp carboxylate groups coordinating in a monodentate fashion to adjacent Co(II) centres, whereas in 2 only one carboxylate group bridges between adjacent Co(II) centres giving rise to a single chiral (10,3)-a net. In 1 and 2 the coordinated water molecules hydrogen bond to the non-coordinated carboxylate oxygen atoms. These interactions give rise to water-carboxylate double helices in , and support the coordination network in 2. Strikingly for a pair of dimorphs the crystal densities of 1 and 2 differ by ca. 0.3 g cm(-3)(1.654 vs. 1.940 g cm(-3), respectively). Compound 3 is isomorphous with 1 and likewise features two chiral interpenetrating nets of quartz topology. In 4, chelating bipyridine nitrogen atoms and four carboxylate oxygen atoms from a total of five adjacent dcbp ligands provide distorted octahedral geometry around Mn(II). The carboxylate groups bridge adjacent Mn(II) centres to produce bis-carboxylato chains which cross-link and generate a 3D network that is perforated with channels. The channels are occupied with disordered DEF molecules. The network topology in 4 is quite different to 1-3 and has a (4.6(2))(4(2).6)(4(3).6(6).8(6)) Schlafli notation. Magnetic susceptibility studies performed on 1, 2, [[Mn(dcbp)].1/2DMF] 5 (DMF = dimethylformamide) and [[Mn(dcbp)].2H(2)O] 6 reveal very weak antiferromagnetic coupling between the metal centres in each case.     

The matched queueing network PH/M/c→oPH/PH/1

We consider a matched queueing network PH/M/coPH/PH/1, where the output of the first subsystem PH/M/c is regarded as one of the inputs of the second subsystem and matched with another input of PH-renewal process, and they are served by a single server with PH-service-distribution. The stationary state probabilities, the distribution of the virtual waiting times in the two subsystems, and the distribution of the total virtual waiting time are derived.The work was supported by the Max-Planck-Gesellschaft, Germany, and the Chinese Academy ofSciences.