Machine learning and quantum theory combine to create new medicines

Studying new medicines and their properties is made that much easier thanks to supercomputers and artificial intelligence. Quantum computing is an ever-increasing reality. Experts believe that AI has democratised the development of medicines and helps to better understand chemistry while questioning certain work methods.

The first medicine created using artificial intelligence was confirmed in 2021. It is to be used for patients suffering from a rare disease, namely, Huntington’s chorea. Creating new medicines, or new functions for already existing ones is no easy matter. How can supercomputing, machine learning and artificial intelligence, or quantum computing discover new medicines?

It is estimated that there are about 1060 drug combinations with known compounds, so trying all of them out is just not feasible every time a disease or ailment has to be studied.

“In silico techniques (meaning computer simulations, an expression that comes from the pseudo-Latin word silico, properly speaking silicon, a basic component of classic computing) enable a pre-filtering of known substances”, explains Paulino Gómez-Puertas, co-ordinator of the Molecular Modelling Team at the Severo Ochoa Molecular Biology Centre (a mixed CSIC-UAM centre) in Madrid, by using supercomputing. “Instead of the millions of initial compounds, 3D computational simulation works like a sieve, selecting dozens of compounds for in vitro (test tube) or in vivo (in animals and even people) testing”.

Moreover, “artificial intelligence and machine learning techniques like deep learning are being used in all stages of drug development”, says Nuria E. Campillo, a scientific researcher at CSIC, as well as a communicator and businesswoman, “which is a very long, expensive and complex process with quite a high failure rate to boot”. The technology involved helps to reduce costs and serves to democratise the access of small and medium sized enterprises (SMEs), as well as start-ups.

What are these new approaches bringing to table? “The main feature of machine learning is that the system is capable of learning socratically. In order words, on the basis of examples and by instructions”, says Alejandro Pazos Sierra, a medical graduate from Complutense University in Madrid (UCM) and holder of a PhD in software engineering from the Polytechnic University of Madrid. He points out two particularly interesting characteristics of this approach: it comes up with solutions to complex problems and has the capacity to learn and adapt.

The power of AI when dealing with small scales

Artificial intelligence, according to Nuria Campillo, enables the designing of new molecules from scratch while also making it possible to outline their properties. “There is enormous room for new molecules [potential combinations]. What we used to do with classic tools was modify existing molecules”. These were trapped in very compact clusters. But now the whole spectrum has been thrown wide open.

Albeit with certain limitations, of course. Paulino Gómez-Puertas tells us that “AI looks for those new drugs that would fit well in the hole (or active site) of a certain enzyme”, so instead of starting out from the enormous databases of compounds we already have, AI designs drugs that do not exist in nature. However, sometimes you come up against the technical problem that the solution in question is impossible or very costly to synthesise. “You’re left wondering if it would have really worked or if it’s what we call in technical-speak, a false positive, because you can’t manufacture it, which precludes it being tried out”.

Another of the problems associated with AI is its black box mechanism. As Alejandro Pazos Sierra points out, quite often “it is unable to explain why it gets the results it does”. There are some fields, such as medical diagnosis for example, where this is critical. Nonetheless, machine learning forges on. It can be used to “identify candidate molecules for drugs, depending on the molecular shape and structure of the target proteins”.

“From the 3D volume standpoint, enzymes [which cause or complicate certain diseases, such as those that render bacteria resistant to antibiotics or facilitate the replication of viruses] are like spheres with a hole. A pocket called the active site”, explains Gómez-Puertas. The approach consists in looking for chemicals to fill this space. Chemicals, therefore, that can deprive the enzyme of the power to impair the work of antibiotics in the case of bacteria and to extend infections in the case of viruses.

Artificial intelligence helps us to better understand chemistry

In 2023, the Fundación BBVA Frontiers of Knowledge Award in biomedicine went to “contributions to the use of artificial intelligence (AI) in accurately predicting the 3D structure of proteins” achieved by the projects, AlphaFold (IA and ML for protein folding) and Rosetta@home (a distributed computing project).

The fact that artificial intelligence and machine learning help us to better understand chemistry, and even to question some of the working methods and limited categories of ‘humans’ (e.g., is a virus a living organism?), is going to be particularly useful when it comes to tackling the appearance of super-bacteria or repositioning drugs.

Repositioning drugs is quite simple to understand, it refers to discovering positive side-effects of drugs already on the market, which means savings on research, testing and approval. The capacity of super-bacteria is a bit harder to grasp: Paulino Gómez-Puertas explains it as follows: “Resistance to antibiotics normally occurs within a segment of DNA called a plasmid, which are found inside bacteria. Bacteria can physically transmit the plasmid between different species”. This represents a big challenge for hospitals. 

Predicting properties

In 2018, Atom2Vec artificial intelligence managed to recreate the periodic table of elements in two hours from a big molecule database. It has taken humans a century to understand how to order this information. According to the Stanford University communiqué, IA can help us “to find new laws of nature”.

At present, we have reached a fascinating point: the predictive capacity of molecular properties that we have tested, thanks to the capacity of AI to learn an enormous amount of data. Predictability is a key factor in pharmacology, “knowing beforehand if a molecule will be toxic or if it is going to be absorbed is very useful”, adds Nuria Campillo, who points out the big cost savings involved. “Changing a molecule after a start has been made on a research project is very expensive”.

Nuria Campillo and her company AItenea Biotech are currently working on designing and optimising molecules and their properties using small sets of data; something that has become possible thanks to the appearance of algorithms that enable training with very few data (it can be likened to how a child learns that a cat is a cat: one or two examples suffice).

The next stage on the search for medicines will come with the aid of quantum computing. As the experience curve cheapens the cost of new methods owing to their massification, their use becomes more widespread. Let’s have a look at some stand-out examples:

• Photovoltaic energy In 1975, it cost over USD 115 to generate 1 watt, whereas by 2020 it cost around USD 0.27. 
• Genome analysis Sequencing the genome cost USD 1 billion in 2000, but by 2015 it could be done at a cost of USD 1,000.  
• Digital storage Storing 1 GB of data in 1980 cost over half a million dollars, a cost which had dropped to USD 0.02 in 2020.

“AI has democratised the development of medicines and has enabled small companies to come on board in their development”, concludes Campillo. When quantum computer algorithms are developed and the quantum computing market stabilises, there is no doubt use of these tools will become widespread among researchers.

US Artificial Intelligence in Healthcare Market (2020 – 2030)

Source: Grand View Research