Molecular graph convolutions: moving beyond fingerprints

Molecular “fingerprints” encoding structural information are the workhorse of cheminformatics and machine learning in drug discovery applications. However, fingerprint representations necessarily emphasize particular aspects of the molecular structure while ignoring others, rather than allowing the model to make data-driven decisions. We describe molecular graph convolutions, a machine learning architecture for learning from undirected graphs, specifically small molecules. Graph convolutions use a simple encoding of the molecular graph—atoms, bonds, distances, etc.—which allows the model to take greater advantage of information in the graph structure. Although graph convolutions do not outperform all fingerprint-based methods, they (along with other graph-based methods) represent a new paradigm in ligand-based virtual screening with exciting opportunities for future improvement.     

Radiometric Sterility Test for Radiopharmaceuticals

A radiometric system originally developed for testing of blood samples obtained from patients suffering from bacteraemia has been adapted for sterility testing of radiopharmaceuticals. Radiometric sterility tests on 202 batches of 19 different radio-pharmaceutical injections were performed and compared with the conventional sterility test method. This technique is more sensitive and rapid than the conventional method.

The radiometric method can be used for injectable radiopharmaceuticals of the radionuclides ³⁵S, ⁵¹Cr, ⁵⁷Co, ⁵⁸Co, ⁵⁹Fe, ⁸²Br, ⁸⁶Rb, ^99mTc, ^113mIn, ¹²³I and ¹⁶⁹Yb. The method is not applicable to iodine-131 labelled radiopharmaceuticals owing to the interference of volatile and radioactive xenon-131m. In the case of neohydrin-203Hg it is necessary to carry out the test after removal of the radioactive material by filtration through a microbe retaining filter.     

Essential oil composition of aerial parts of Cyclospermum leptophyllum (Pers.) Sprague ex Britton and P. Wilson

The essential oil of Cyclospermum leptophyllum (Pers.) Sprague ex Britton and P. Wilson syn. Apium leptophyllum (Pers.) F. Muell. Ex Benth., family Apiaceae (Umbellifereae), was analysed by GC and GC-MS. The essential oil showed the presence of thymohydroquinone dimethyl ether (50.7%), thymol methyl ether (11.2%), γ-terpinene (10.4%), p-cymene (9.5%) and carvacrol methyl ether (5.9%), along with trans-ocimene, β-pinene, germacrene D, cumin aldehyde, α-humulene, β-myrcene, sabinene, limonene, β-phellandrene, caryophyllene oxide, α-pinene and α-thujene as minor constituents. To the best of our knowledge, this is the first study of the essential oil composition of C. leptophyllum collected from India.     

Error bounds on the SCISSORS approximation method

The SCISSORS method for approximating chemical similarities has shown excellent empirical performance on a number of real-world chemical data sets but lacks theoretically proven bounds on its worst-case error performance. This paper first proves reductions showing SCISSORS to be equivalent to two previous kernel methods: kernel principal components analysis and the rank-k Nystro?m approximation of a Gram matrix. These reductions allow the use of generalization bounds on these techniques to show that the expected error in SCISSORS approximations of molecular similarity kernels is bounded in expected pairwise inner product error, in matrix 2-norm and Frobenius norm for full kernel matrix approximations and in root-mean-square deviation for approximated matrices. Finally, we show that the actual performance of SCISSORS is significantly better than these worst-case bounds, indicating that chemical space is well-structured for chemical sampling algorithms.     

Comparison of denitrification between Paracoccus sp. and Diaphorobacter sp

Denitrification was compared between Paracoccus sp. and Diaphorobacter sp. in this study, both of which were isolated from activated sludge of a denitrifying reactor. Denitrification of both isolates showed contrasting patterns, where Diaphorobacter sp. showed accumulation of nitrite in the medium while Paracoccus sp. showed no accumulation. The nitrate reduction rate was 1.5 times more than the nitrite reduction in Diaphorobacter sp., as analyzed by the resting state denitrification kinetics. Increasing the nitrate concentration in the medium increased the nitrite accumulation in Diaphorobacter sp., but not in Paracoccus sp., indicating a branched electron transfer during denitrification. Diaphorobacter sp. was unable to denitrify efficiently at high nitrate concentrations from 1 M, but Paracoccus sp. could denitrify even up to 2 M nitrate. Paracoccus sp. was found to be an efficient denitrifier with insignificant amounts of nitrite accumulation, and it could also denitrify high amounts of nitrate up to 2 M. Efficient denitrification without accumulation of intermediates like nitrite is desirable in the removal of high nitrates from wastewaters. Paracoccus sp. is shown to suffice this demand and could be a potential organism to remove high nitrates effectively.     

Acylation of Grignard reagents mediated by N-methylpyrrolidone: a remarkable selectivity for the synthesis of ketones

An efficient user-friendly method of acylation of Grignard reagents to selectively synthesize ketones is presented, which is assisted by simple amides such as NMP, or DMF. The present chemoselective method tolerates a variety of functional groups such as ketone, ester, nitrile and other functional groups.     

Slow Unfolded-State Structuring in Acyl-CoA Binding Protein Folding Revealed by Simulation and Experiment

Protein folding is a fundamental process in biology, key to understanding many human diseases. Experimentally, proteins often appear to fold via simple two- or three-state mechanisms involving mainly native-state interactions, yet recent network models built from atomistic simulations of small proteins suggest the existence of many possible metastable states and folding pathways. We reconcile these two pictures in a combined experimental and simulation study of acyl-coenzyme A binding protein (ACBP), a two-state folder (folding time ～10 ms) exhibiting residual unfolded-state structure, and a putative early folding intermediate. Using single-molecule FRET in conjunction with side-chain mutagenesis, we first demonstrate that the denatured state of ACBP at near-zero denaturant is unusually compact and enriched in long-range structure that can be perturbed by discrete hydrophobic core mutations. We then employ ultrafast laminar-flow mixing experiments to study the folding kinetics of ACBP on the microsecond time scale. These studies, along with Trp-Cys quenching measurements of unfolded-state dynamics, suggest that unfolded-state structure forms on a surprisingly slow (～100 μs) time scale, and that sequence mutations strikingly perturb both time-resolved and equilibrium smFRET measurements in a similar way. A Markov state model (MSM) of the ACBP folding reaction, constructed from over 30 ms of molecular dynamics trajectory data, predicts a complex network of metastable stables, residual unfolded-state structure, and kinetics consistent with experiment but no well-defined intermediate preceding the main folding barrier. Taken together, these experimental and simulation results suggest that the previously characterized fast kinetic phase is not due to formation of a barrier-limited intermediate but rather to a more heterogeneous and slow acquisition of unfolded-state structure.     

Some recent advances in the theory of homogeneous isotropic turbulence

We review some advances in the theory of homogeneous, isotropic turbulence. Our emphasis is on the new insights that have been gained from recent numerical studies of the three-dimensional Navier Stokes equation and simpler shell models for turbulence. In particular, we examine the status of multiscaling corrections to Kolmogorov scaling, extended self similarity, generalized extended self similarity, and non-Gaussian probability distributions for velocity differences and related quantities. We recount our recent proposal of a wave-vector-space version of generalized extended self similarity and show how it allows us to explore an intriguing and apparently universal crossover from inertial- to dissipation-range asymptotics.     

Instrumentation for PSD-based neutron diffractometers at Dhruva reactor

Linear position sensitive detectors (PSDs) are widely used to configure neutron diffractometers and other instruments. Necessary front-end electronics and a data acquisition system [1] is developed to cater to such instruments built around the Dhruva research reactor in BARC. These include three diffractometers with multiple PSDs and four with single PSD. The front-end electronics consists of high voltage units, preamplifiers [2],shaping amplifiers, ratio ADCs (RDC) [3]. The data acquisition system consists of an interface card and software. Commercially available hardware like temperature controller or stepper motor controller connected over GPIB or RS232 are also integrated in the data acquisition system. The data acquisition is automated so that it can continue unattended for control parameter like temperature, thus enabling optimum utilization of available beam time. The instrumentation is scalable and can be easily configured for various instrumental requirements. The front-end electronics and the data acquisition system are described here.