Collaborating Authors

A tiny revolution? Three scientists win Nobel Prize for molecule machines

Christian Science Monitor | Science

Alfred Nobel wanted the prizes that bear his name to recognize achievements that offered the "greatest benefit to mankind." The world's tiniest machines -- celebrated in this year's chemistry prize -- may revolutionize daily life. The Royal Swedish Academy of Sciences on Wednesday awarded the final Nobel prize in sciences for 2016. The 8 million kronor ( 930,000) chemistry prize went to Jean-Pierre Sauvage of France, Sir Fraser Stoddart of Britain, and Bernard "Ben" Feringa of the Netherlands. The scientists were recognized for their breakthroughs on molecular machines, which began with Dr. Sauvage linking two ring-shaped molecules in 1983.

DeepSIBA: Chemical Structure-based Inference of Biological Alterations Machine Learning

Predicting whether a chemical structure shares a desired biological effect can have a significant impact for in-silico compound screening in early drug discovery. In this study, we developed a deep learning model where compound structures are represented as graphs and then linked to their biological footprint. To make this complex problem computationally tractable, compound differences were mapped to biological effect alterations using Siamese Graph Convolutional Neural Networks. The proposed model was able to learn new representations from chemical structures and identify structurally dissimilar compounds that affect similar biological processes with high precision. Additionally, by utilizing deep ensembles to estimate uncertainty, we were able to provide reliable and accurate predictions for chemical structures that are very different from the ones used during training. Finally, we present a novel inference approach, where the trained models are used to estimate the signaling pathways affected by a compound perturbation in a specific cell line, using only its chemical structure as input. As a use case, this approach was used to infer signaling pathways affected by FDA-approved anticancer drugs.

Generating equilibrium molecules with deep neural networks Machine Learning

Discovery of atomistic systems with desirable properties is a major challenge in chemistry and material science. Here we introduce a novel, autoregressive, convolutional deep neural network architecture that generates molecular equilibrium structures by sequentially placing atoms in three-dimensional space. The model estimates the joint probability over molecular configurations with tractable conditional probabilities which only depend on distances between atoms and their nuclear charges. It combines concepts from state-of-the-art atomistic neural networks with auto-regressive generative models for images and speech. We demonstrate that the architecture is capable of generating molecules close to equilibrium for constitutional isomers of C$_7$O$_2$H$_{10}$.

A Generative Model for Molecular Distance Geometry Machine Learning

Computing equilibrium states for many-body systems, such as molecules, is a long-standing challenge. In the absence of methods for generating statistically independent samples, great computational effort is invested in simulating these systems using, for example, Markov chain Monte Carlo. We present a probabilistic model that generates such samples for molecules from their graph representations. Our model learns a low-dimensional manifold that preserves the geometry of local atomic neighborhoods through a principled learning representation that is based on Euclidean distance geometry. We create a new dataset for molecular conformation generation with which we show experimentally that our generative model achieves state-of-the-art accuracy. Finally, we show how to use our model as a proposal distribution in an importance sampling scheme to compute molecular properties.

Tomorrow Edition - Interview with Mathematician, Medical Scientist, and Quantum Computation Specialist Dr. Joseph Geraci


Dr. Geraci's efforts mainly concern precision medicine, using mathematical and computational methods to construct models of disease that go beyond classical top-down clinical definitions. After completing postdocs in oncology, biological psychiatry, and artificial intelligence he created NetraMark Corp where he has been developing novel technologies that aid in the understanding of our molecular and brain circuitry in addition to novel machine learning algorithms specialized to help understand complex patient populations. He is also a professor of Molecular Medicine at Queen's University in Ontario. Dr. Geraci has a strong interest in advancing the mathematical methods being employed in the study of our molecular circuitry (protein, microRNA, mRNA), the analysis of brain MRIs, and machine learning that can use variables that are beginning to emerge due to our interaction with technologies like fitness watches and smart buildings. A major interest of his is an ongoing project involving translating the vast amount of genetic and proteomic patient data, coupled with our current knowledge of our molecular circuitry, into a scoring scheme that can reveal potential new drug targets.