Over the past decades, chemical sciences have helped us develop a large number of materials, ranging from polymers to thermoplastic, having applications in photographic films, paint, and clothing industries. Nevertheless, the challenges confronting us, in the form of pandemics and climate change, have forced us to look out for faster means to develop new materials to meet the demands and issues of recent times. It has become the need of the hour to develop and design new medicines at an unmatched speed and scale.
In sync with this, adding yet another feather to its technological know-how, IBM has launched – Molecule Generation Experience (MolGX), a cloud-based, AI-driven platform to design novel molecular designs for the development of new materials.
Typically, it takes up to 10 years and $10–100 million in funding to discover a new molecule. As a matter of fact, the vastness of chemical sciences hinders the possibility of tracking even a smaller fraction of material structure available around, capable of providing enormous solutions to mankind.
However, the MolGX platform, with its capability to decipher an infinite number of patterns in the shape of the molecules — the very basis of deciding the material properties, thereby reducing the enormous amount of time scientists and researchers put into developing new materials. It will accelerate the development by 10 to 100 times – compared with the usual circumstances.
The tech behind
Based on the idea of reverse designing, AI is used to draw realistic images of landscapes or portraits of people that don’t even exist – termed as generative models. IBM’s MolGX, equipped with this technology, can analyse desired chemical properties such as “solubility in water” and “heatability” of the materials provided to generate new molecules.
The tool has to perform three simple steps to execute the same — observing and selecting a dataset, training the AI model to foresee molecules’ chemical properties within set parameters, and designing the molecular structure in the end – based on the model created. MolGX is a novel approach to create new materials because it eliminates production time and creates new materials that are not constrained by preconceived notions.
Firstly, to facilitate this, IBM Deep Search extracts the structured data comprising structures and characteristics of molecules, which is further complemented by AI-accelerated simulation and integrated into MolGX to train the generative artificial intelligence. Secondly, the Simplified Molecular-input line-entry system (SMILES) – supported by MolGX, represents a line drawing of a molecule in the form of a sequence of characters — BrCCOC1OCCCC1.
The generated molecules by MolGX are then passed to IBM RXN for Chemistry, where complex molecules are reduced into a simpler structure helping researchers to identify the starting material from which a chemical synthesis can be developed, and IBM RoboRXN that automates synthesis procedures by converting chemical recipes to machine-readable instructions.
Advantages of MolGX
The tech giant has deployed its novel AI-driven molecular inverse-design platform to develop and produce photoacid generators, one of the important components in electronics manufacturing. One of the major advantages it brings to material research and development is its use of an algorithm-based encoding and structure generation process rather than a data-driven one, eliminating the need for large dataset pre-training and the associated high training costs.
In addition to the above, human chemists can completely understand the space and structure generation process, making it simple to customise chemical insights about molecular structures. Alongside, MolGX comes with a versatile and intuitive workflow and a simple user interface.
The MolGX technical version is now being used as a service at NAGASE, a materials manufacturing firm, where it helps in inverse-designing sugar and dye molecules ten times faster than human chemists.
Summing up the development
Apart from IBM, companies like Ketobix are also developing AI technologies to discover new materials. The company has launched the world’s first self-driving lab for materials discovery powered by artificial intelligence and robotics, which has now stepped up its quest for new molecules and materials, as well as their discovery, application, and production.
Experts believe that new technologies, especially those utilising artificial intelligence and hybrid clouds, like IBM’s MolGX and Ketobix, will have a more significant role in the coming times. The platforms provide humanity with an unmatchable speed and scale to discover new materials, useful to address global issues with a positive impact on our ecosystem. Data-processing technologies and artificial intelligence can drastically cut the time required and are cost-efficient, which needs to be groomed-up with a good investment.