decoding diseased cells
Biontech researches cancer like others research the weather
1/22/2023 3:59 p.m
Artificial intelligence aims to accelerate Biontech’s cancer research. In the “Stern” podcast “A new medicine – the Biontech story”, company boss Sahin explains the role played by algorithms in the development of tailor-made vaccines.
It is the largest takeover in Biontech’s young history: the Mainz-based vaccine manufacturer buys the British company for more than 400 million euros InstaDeepwhose core business is developing artificial intelligence (AI): programs that independently find solutions to problems that a normal computer fails to solve.
The fastest vaccine development of all time saves millions of lives and comes from a previously unknown company in Mainz. The podcast “A new medicine – the Biontech story” tells how the founding couple Uğur Şahin and Özlem Türeci made the breakthrough after three decades of research with an mRNA vaccine during the corona pandemic and what they are planning next: therapies against HIV, tuberculosis and cancer – and maybe even one to stop aging. Now at RTL+ Music.
Recently, artificial intelligence has often been reported in connection with so-called chatbots. It’s now working Programs like ChatGPTto independently write texts of such a quality that it is often difficult for a person to unmask them as the work of a computer.
But AI is suitable for many other applications, including medical ones. Because powerful computers are able to discover patterns in large data collections and to draw correct conclusions from them. This plays a crucial role, for example, in automatic image recognition for self-driving cars. However, an AI can also help in the development of new drugs by creating the molecular design of active ingredients – for example in the case of individual vaccines against cancer.
Cancer is a complicated opponent
Uğur Şahin and Özlem Türeci founded Biontech in 2008 to develop such vaccines for the treatment of tumors. Even then, the messenger molecule mRNA was the basis for these innovative therapies. The technology was used on a large scale for the first time with the corona vaccine.
However, cancer cells are a much more complex opponent than corona viruses because they differ greatly from one another. Different types of tumors in different patients – the cancer cells from two patients may look alike, but they never look the same. Hence the obvious thought that therapy must be tailor-made.
enemy in your own body
An effective mRNA vaccine against must be able to teach the immune system what is foreign about a cancer cell and thus distinguish it from a healthy cell in the body. In this way, the immune system should learn which cells to attack and which not. There is still no cancer vaccine on the market, although research has been going on for decades. The project is tricky because cancer cells – unlike Sars-CoV-2, flu viruses or transplanted organs – come from your own body and are therefore very similar to body cells.
Companies like Biontech are therefore looking for characteristics in the genome of patients’ cancer cells that distinguish them from healthy cells. These can be gene sections of certain proteins on the surface of cancer cells, for example. These gene sequences then serve as the basis for the mRNA vaccination.
Hunt for promising traits
A major problem, however, is that not all differences in the genome of cancer cells and healthy cells are suitable as a basis for a vaccine. Researchers and physicians have to choose the right mutation: the one that the immune system will later respond to.
At this point, Şahin and Türeci hope for decisive support from artificial intelligence. The AI is fed with data from clinical studies and genome databases and can then – according to the plan – use algorithms to recognize and select the most promising characteristics in the genome of the cancer cells. This should eliminate lengthy laboratory tests and bring effective cancer vaccines tailored to individual patients within reach.
Like the weather forecast
The right algorithms are crucial here, as is the data basis. The better and more numerous the data, the more accurately an AI can make its predictions. Biontech founder Şahin compares the process with a weather forecast. They are based on similar algorithms as ten years ago, explains Şahin in the “Stern” podcast “A new medicine – the Biontech story”. “But the data is so good and the computing capacity so strong that we can now work much more precisely and precisely.”
The Mainz-based company has been working with InstaDeep for several years, and buying the AI manufacturer is a logical step. By 2030 Biontech would like to have already treated several thousand people with cancer therapies developed in-house. “In the early stages we can defeat cancer,” Şahin is convinced in the podcast. “Our goal must be to go from 60 to 99 percent.”
Extensive, advanced tumors become more complicated. You have to be realistic, says the Biontech founder. “It’s about tens of billions of tumor cells, all of which are different.” Here, through the interaction of different methods, one will most likely ensure “that the tumor is controlled for as long as possible”.