Toxicity prediction of chemical compounds is a grand challenge. Lately, it achieved significant progress in accuracy but using a huge set of features, implementing a complex blackbox technique such as a deep neural network, and exploiting enormous computational resources. In this paper, we strongly argue for the models and methods that are simple in machine learning characteristics, efficient in computing resource usage, and powerful to achieve very high accuracy levels. To demonstrate this, we develop a single task-based chemical toxicity prediction framework using only 2D features that are less compute intensive. We effectively use a decision tree to obtain an optimum number of features from a collection of thousands of them. We use a shallow neural network and jointly optimize it with decision tree taking both network parameters and input features into account. Our model needs only a minute on a single CPU for its training while existing methods using deep neural networks need about 10 min on NVidia Tesla K40 GPU. However, we obtain similar or better performance on several toxicity benchmark tasks. We also develop a cumulative feature ranking method which enables us to identify features that can help chemists perform prescreening of toxic compounds effectively.

No abstract

We report a structure-based approach to design peptides that can bind to aggregation-prone, partially folded intermediates (PFI) of insulin, thereby inhibiting early stages of aggregation nucleation. We account for the important role of the modular architecture of protein–protein binding interfaces and tertiary structure heterogeneity of the PFIs in the design of peptide inhibitors. The determination of association hotspots revealed that two interface segments are required to capture majority contribution to insulin homodimer binding energy. The selection of peptides that will have a high probability to inhibit insulin self-association was done on the basis of similarity in binding interface coverage of PFI residues in the peptide–PFI complex and the native–PFI dimer. Data on aggregate growth rate and secondary structure for formulations incubated under amyloidogenic conditions show that designed peptides inhibit insulin aggregation in a concentration-dependent manner. The mechanism of aggregation inhibition was probed by determining the enthalpy of peptide–insulin binding and peptide micellization using isothermal titration calorimetry. Finally, the effect of designed peptides on insulin activity was quantified using a spectrophotometric assay for glucose uptake by HepG2 cells.