BiopharmaTrend.com

Generative models have become one of the hottest areas in de-novo molecular design over just several years, basically revolutionizing our perception of what can be done with artificial intelligence in this area. One important aspect of generative models is that they can produce new quality hit molecules using combined data from various experimental and theoretical sources -- and output results rapidly. 

One notable drug discovery startup betting on deep learning and generative models for innovative drug design is Vancouver-based Variational AI.

Tzager is an A.I. agent built for biomedicine research, drug discovery and personalized medicine, with the main features being Biochemical Analysis, Predictor Research/Models and Literature Review/Management. The difference with Tzager is that it is not just another deep learning algorithm trained to solve very specific problems, but the intelligence system with its own framework based on Causal Equations and Bayesian Networks.

As we enter a new decade, our belief in the the impact of Artificial intelligence (AI) is only getting stronger. Supporting the industry to drive the right drug to the right patient at speed is a huge responsibility that we take very seriously. Towards the end of the last decade we have seen great progress made within life sciences and the use of AI, but moving into 2020 the spotlight on commercial teams and gaining competitive advantage with AI will intensify.

Chemical Data Has Problems

The state of data access, quality and dissemination in Chemistry is extremely poor - so poor that it is blocking advances in machine learning (ML) and artificial intelligence (AI), and also impeding research and development in traditional methods. The recent surge in AI skepticism is a direct consequence of years of over-hype and promises based on precarious data. Over-the-top expectation were offered without enough consideration for the data quality and volume required to train fancy algorithms. The old adage “^&$% in, ^&$% out” holds true (we can say ‘crap’ right?). This opinion is in line with recent statements by the CEO of Novartis, for example, who runs the second largest pharmaceutical company in the world, lamenting the difficulty in accessing quality datasets to make AI effective.

Computer-aided drug design (CADD) is a central part of so-called “rational drug design”, pioneered in the last century by companies like Vertex. Over the last decades, CADD had great influence on the way new therapeutics are discovered, however, it also showed limitations due to modest accuracy of computational tools.  

The majority of software tools used for computational chemistry and biology rely on molecular mechanics -- a simplified representation of molecules, essentially reducing them down to “balls and sticks”: atoms and bonds between them. In this way it is easier to compute, but accuracy suffers greatly.

In order to gain adequate accuracy, one has to account for the electronic behavior of atoms and molecules, i.e. consider subatomic particles -- electrons and protons. This is what quantum mechanical (QM) methods are all about -- and the theory is not new, dating back to the early decades of the 20th century.