System Bits: Feb. 14

Potential anticancer drugs selected by neural network
Moscow Institute of Physics and Technology researchers along with Mail.Ru Group, and Insilico Medicine have applied a generative neural network to create new pharmaceutical medicines with certain desired characteristics.

A generative adversarial network (GAN) was developed and trained to “invent” new molecular structures in order to dramatically reduce the time and cost of searching for substances with potential medicinal properties. In this case, the researchers intend to use these technologies in the search for new medications within various areas from oncology to CVDs and even anti-infectives.

Andrei Kazennov, one of the authors of the study and an MIPT postgraduate who works at Insilico Medicine, said, “We’ve created a neuronal network of the reproductive type, i.e. capable of producing objects similar to what it was trained on. We ultimately taught this network model to create new fingerprints based on specified properties.”
(Source: Moscow Institute of Physics and Technology)

The researchers explained that currently, the inorganic molecule base contains hundreds of millions of substances, and only a small fraction of them are used in medicinal drugs. The pharmacological methods of making drugs generally have a hereditary nature. For example, they pointed to the fact that pharmacologists might continue to research aspirin that has already been in use for many years, perhaps adding something into the compound to reduce side effects or increase efficiency, while the substance still remains the same. Earlier this year, the scientists at Insilico Medicine demonstrated that it is possible to substantially narrow the search using deep neural networks. But now they have focused on the much more challenging question of whether there is a chance to create conceptually new molecules with medicinal properties using the novel flavor of deep neural networks trained on millions of molecular structures.

In essence, the Generative Adversarial Autoencoder (AAE) architecture, an extension of Generative Adversarial Networks, was taken as the basis, and compounds with known medicinal properties and efficient concentrations were used to train the system. Information on these types of compounds was input into the network, which was then adjusted so that the same data was acquired in the output.

Read more about their developments here.

Simulating ransomware attacks
Meant to highlight vulnerabilities in industrial control systems used to operate industrial facilities such as manufacturing plants, water and wastewater treatment facilities, and building management systems for controlling escalators, elevators and HVAC systems, Georgia Institute of Technology researchers have developed a new form of ransomware that was able to take over control of a simulated water treatment plant. After gaining access, the researchers were able to command programmable logic controllers (PLCs) to shut valves, increase the amount of chlorine added to water, and display false readings.

Georgia Tech researchers have developed a new form of ransomware that can take over control of a simulated water treatment plant. The simulated attack was designed to highlight vulnerabilities in the control systems used to operate industrial facilities. Shown are (left) Raheem Beyah, associate chair in the Georgia Tech School of Electrical and Computer Engineering, and David Formby, a Georgia Tech Ph.D. student. (Source: Georgia Tech) 

While the researchers say no real ransomware attacks have been publicly reported on the process control components of industrial control systems, the attacks have become a significant problem for patient data in hospitals and customer data in businesses as attackers gain access to these systems and encrypt the data, demanding a ransom to provide the encryption key that allows the data to be used again. They believe it’s only a matter of time before critical industrial systems are compromised and held for ransom.

Neutrino viewer app
It used to be that observing fundamental particles was reserved for scientists with complex equipment but now anyone can explore the world of particles from their phone, thanks to a new, and free smartphone app developed by researchers at Oxford University.

VENu is aimed at supporting would-be physicists to see neutrino activity, and to even try to catch them themselves. According to Oxford, the app is made up of data gathered by scientists from the Microboone experiment, launched to detect and understand neutrinos, which are subatomic, almost weightless particles that only very rarely interact. They are notoriously difficult to capture but state-of-the-art detectors, like Microboone, in the USA, are now recording neutrino interactions. The footage captured enables scientists to understand more of this elusive and puzzling particle. They reminded that neutrinos are considered a fundamental building block of matter, and are fascinating to scientists. They carry no electric charge and can travel through the universe almost entirely unaffected by natural forces, and are therefore very difficult to detect.

VENu works with Google Cardboard and is designed to exhibit both virtual and augmented reality features. The personal virtual reality viewer allows users to understand the many complexities and intricacies of the Microboone experiment and to learn more about neutrinos.

There are game features for brain teasing challenges, and putting them in the mind-set of professional particle physicists including simulating neutrino interactions against a cosmic ray background, similar to the way neutrino physicists run their analysis.