The molecular design and exploration industry will continue to evolve at an accelerated pace in 2019, with technology, automation and AI playing an ever-increasing role. As we wrote this piece and “ The Year In Molecular Design & Exploration ”, we noticed a trend — the application of technology towards the improvement of scientific experimentation and drug design.
Let’s take a look at some of the technologies and how they were applied in 2018, the discussion had along the way, and what you might expect in 2019.
Alán Aspuru-Guzik of Harvard University created a free software package called ChemOS. The software, which he developed, allows scientists to apply automation in their work. “If you look at the chemistry lab of the 16th century or even the 21st century, you will see the same thing,” Aspuru-Guzik said. “Nothing has changed really. If we really want to rethink discovery, we need to rethink the laboratory.”
He also showed that ChemOS could be used remotely, experimenting with the software in the laboratory of organic chemist Jason E. Hein at the University of British Columbia from Harvard.
Throughout 2019 we’ll see chemists continue to harness software of all types, including artificial intelligence, which helps scientists do things like predict chemical reactions.
In 2018 we saw the fruits of such labor. Efforts by Abigail Doyle , a group of researchers at the A. Barton Hepburn Professor of Chemistry at Princeton University and Spencer Dreher of Merck Research Laboratories found a way to predict reaction yields accurately by using machine learning . (Their software is now available to other chemists)
“The software that we developed is designed to accommodate any reaction or substrate type,” said Doyle. “The idea was to let someone apply this tool and hopefully build on it with other reactions.”
Chemists can use the software to identify high-yielding combinations of chemicals and substrates more affordably and efficiently. “We hope this will be a valuable tool in expediting the synthesis of new medicines,” said Derek Ahneman, who completed his chemistry Ph.D. under Doyle lab in 2017, and now works for IBM.
“Many of these machine learning algorithms have been around for quite some time,” said Jesús Estrada , a graduate student in Doyle’s lab . “However, within the synthetic organic chemistry community, we really haven’t tapped into the exciting opportunities that machine learning offers.”
Computational chemistry will also continue to help address challenges in designing functional materials, as proven by The University of Marburg’s Dr. Ralf Tonner, who is working with collaborator Lisa Pecher. At the High-Performance Computing Center Stuttgart (HLRS), one of three German national supercomputing centers at the Gauss Centre for Supercomputing, Tonner models phenomena at the atomic and subatomic scale.
This helps scientists understand how things such as molecular structure, electronic properties, chemical bonding, and interactions among atoms affect the behavior of the material.
“When you study how, for example, a molecule adsorbs on a surface other scientists will often describe that phenomenon with methods from physics, solid-state theory, or band structures,” explained Tonner. “We think it can also be very helpful to ask, how would a chemist look at what’s happening here?” Tonner says that in 2019 he might study whether understanding chemical reactions can offer new and useful insights.
Molecular modeling will continue to be a valuable and essential tool in medicinal chemistry for the drug design process. Molecular modeling can be used in drug design, computational biology, computational chemistry, and materials science. It involves a wide range of computerized techniques based on theoretical chemistry methods and experimental data to analyze the biological and molecular property.
AI will continue to drive advancements in drug discovery and design. Merck, for instance, has taken the lead in implementing AI-based solutions in drug discovery. It’s partnered with Numerate, an AI-based company leveraging algorithms and cloud computing to transform the drug design process.
Merck is using Numerate’s computer-based drug design technology to develop novel small molecule drug leads for undisclosed cardiovascular disease target. Further, Merck and Atomwise, which uses deep learning technology for the discovery of novel small molecules, are using AI-based technology to scan existing medicines that could be redesigned.
Pharmaceutical companies partnering with AI-focused companies is an emerging trend, with Celgene having partnered with GNS Healthcare and GSK entered a drug discovery collaboration with UK-based AI-driven startup Exscientia, and Pfizer entered a collaboration with IBM Watson for immuno-oncology drug discovery research, and the list goes on.
AI in drug discovery is made possible by deep learning, a field of machine learning that is built using artificial neural networks that model the way neurons in the human brain communicate. Deep learning trains systems to analyze large sets of chemical and biological data to identify drug candidates successfully and quickly.
With software and technology playing a greater role in scientific exploration, ethical discussions will continue to be a hot topic. On a 2018 episode of “The Future of Everything” host Russ Altman, MD, PhD, spoke with Stanford biomedical ethicist David Magnus, PhD, about how AI could impact the future of ethics and health care.
Dr. Magnus fears potential hidden biases in machine algorithms that could influence the AI’s clinical advice. Let’s say an algorithm guides decisions on which patients go to the intensive care unit. It could be motivated by an individual’s insurance status and put the hospital’s finances above patient care.
AI systems could be designed to perform unethically or be biased based on its input data. If a machine algorithm receives data demonstrating that developmentally delayed children respond poorly to organ transplants, such biases could be continuously reinforced and validated in the medical care system. AI could also diminish the physician’s autonomy while leaving them perplexed as to how to handle and implement the information AI gives them.
With all that said, AI will continue to help physicians simplify the field of genetics, which is comprised of so many data points it can be challenging for physicians to comprehend.
“The potential for using algorithms to really help make decisions under a lot of uncertainty where clinicians don’t have the expertise to do this is really going to be important,” said Dr. Magnus.
What do you think about tech’s expanding role in science in 2019 ? Let us know in the comments.