Just a year ago, biologists first created a completely new antibody, relying not on conventional laboratory methods, but on artificial intelligence calculations. At the time, these molecules were demonstrations of possibilities than actual drug candidates. Now the situation is changing: several scientific groups and companies have announced that for the first time AI-developed antibodies have properties sufficient to enter clinical trials. The results are published in Nature.

Why AI is so important for antibody development

The traditional way of searching for antibodies is reminiscent of trying thousands of keys in the hope of finding the one that fits the right lock. Scientists select random molecules and test whether they can bind strongly to a target—for example, a protein from a virus or a receptor involved in a disease. It happens that there are simply no suitable options.

a و the و of – تفاصيل مهمة

AI is changing the approach. Now researchers can pre-specify the desired target – a section of the protein to which the antibody should “stick” with maximum precision. The algorithm selects a structure that is suitable for this task at the level of individual atoms.

“AI-guided design promises atomic precision,” notes Serge Biswas, head of Nabla Bio.

Such precision makes it possible to create molecules that are practically never found in nature, but at the same time have the necessary medicinal properties.

Breakthrough in modeling complex antibody regions

The main difficulty in developing antibodies using AI was that the key regions of these molecules – the flexible loops by which the antibody “recognizes” its target – are very difficult to predict. They are mobile and often take different forms.

the و are و in – تفاصيل مهمة

Updated models, including an improved version of AlphaFold, can model such structures much accurately. This was a turning point.

In October, the Gabriele Corso group presented a modelBoltzGenwhich made it possible to create “nanobodies” – miniature antibodies similar to those found in sharks and camels. Over several dozen attempts, researchers have obtained molecules that can effectively bind to proteins involved in cancer, viral infections and other diseases.

In parallel, teams from Stanford, the Arc Institute, and David Baker’s laboratory reported their progress. All of them showed that AI can now design not only nanobodies, but also full-sized antibodies – the same ones that pharmaceutical companies use.

The first molecules approaching drug standards

Nabla and Chai Discovery have already announced that they have created antibodies that behave like commercial drugs in laboratory tests. Some designs have been able to recognize targets that are considered almost inaccessible to classical methods. Among them are GPCRs, receptors that play a key role in cell signaling.

that و to و in – تفاصيل مهمة

Some AI-developed antibodies have shown important drug properties: stability in solution, high concentration without aggregation, and precise recognition of the desired target.

Scientists, however, urge caution. The sequences of these antibodies have not yet been published, and independent researchers cannot verify the reported results. But the very fact that such statements are appearing shows how quickly the field is moving.

When will human trials begin?

The first clinical trials of drugs created using AI have already begun. Generate Biomedicine launched a clinical trial of an antibody against severe asthma – however, the algorithms in this case did not create a molecule from scratch, but improved an existing one, increasing binding strength and stability.

For now, questions remain. The most important thing is how the body will perceive a completely new molecule created by AI: will it be safe, will it not cause unwanted immune reactions. According to Biswas, there is no indication that such antibodies are different from traditional ones, but additional tests will be needed before testing.

the و will و The – تفاصيل مهمة

Additionally, researchers do not yet fully understand which targets are best suited for AI-engineered antibodies.Perhaps algorithms will help “hack” those proteins against which scientists have been unable to create effective drugs for decades.

A new era in biomedicine

Creating antibodies on demand – almost like writing a program – opens the way to drugs that were previously thought impossible.Now that the generation button actually works, scientists can focus on the toughest challenges: drug penetration into the brain, combination targets, or rare diseases where traditional methods fail.

“Now that we have the ability to generate antibodies at the touch of a button, we can spend time on these cutting-edge problems,” Biswas emphasizes.

AI doesn’t just speed up drug development. It changes the very logic of the search – and, perhaps, in the coming years, the first drugs completely created by algorithms will appear in clinics.

the و in و on – تفاصيل مهمة