How AI can help solve the problem of treating rare diseases

Modern biotech has the tools to modify genes and develop drugs, yet thousands of rare diseases remain untreatable. According to the executives of Insilico Medicine and GenEditBio, what has been missing for years has been finding enough smart people to continue the work. AI, he says, is increasing the power that allows scientists to solve problems that industry has left untouched.

Speaking this week at Web Summit Qatar, Insilico’s president, Alex Aliper, described his company’s mission to create “medical intelligence.” Insilico recently launched “MMA Gym” which aims to train people in major languages, such as ChatGPT and Gemini, to work effectively as professionals.

The goal is to build multiple, multi-species models that, Aliper says, can handle multiple drug discovery tasks at once with greater accuracy than humans.

“We need this technology to increase the productivity of our pharmaceutical industry and address the lack of work and talent in this area, because there are still thousands of diseases without treatment, without treatment methods, and there are thousands of rare diseases that are neglected,” Aliper said in an interview with TechCrunch. “So we need very smart systems to deal with this problem.”

The Insilico platform integrates biology, chemistry, and medicine to generate hypotheses about disease targets and target molecules. Using steps that once required chemists and biologists, Insilico says it can analyze large manufacturing sites, select the best drug candidates, and repurpose existing drugs — all at a much lower cost and time.

For example, the company recently used its AI models to determine whether an existing drug could also be used to treat ALS, a rare mental illness.

But the problem of work does not end with access to drugs. Even if AI can identify potential treatments or cures, many diseases require intervention at a more fundamental biological level.

GenEditBio is part of the “second wave” of CRISPR gene editing, in which the process moves away from changing cells outside the body (ex vivo) and towards precise movements inside the body (in vivo). The company’s goal is to create a one-size-fits-all injection directly into the affected tissue.

“We’ve developed an ePDV, or protein delivery vehicle, and it’s like a microparticle,” GenEditBio founder and CEO Tian Zhu told TechCrunch. “We learn from the environment and use AI machine learning techniques to mine the environment and find out which viruses interact with which types of tissue.”

The “biome” Zhu is referring to is GenEditBio’s vast library of thousands of unique, non-viral, lipid-free polymer nanoparticles — delivery vehicles designed to carry gene editing tools into specific cells.

The company says its NanoGalaxy platform uses AI to analyze data and determine how fluid structures interact with tissue targets (such as the eye, liver, or nervous system). The AI ​​then predicts which tweaks to the chemistry of the delivery vehicle will help deliver the goods without triggering the immune system.

GenEditBio tests its ePDVs in vivo in wet labs, and the results are fed back into the AI ​​to refine its predictive models.

Efficient, tissue-specific delivery is essential for gene transfer in vivo, Zhu says. He says that his company’s strategy reduces the cost of goods and establishes a strategy that has been difficult to expand.

“It’s like finding an off-the-shelf drug (that works) for multiple patients, making the drug affordable and available to patients around the world,” Zhu said.

His company recently received FDA approval to start trials of CRISPR treatment for corneal dystrophy.

Coping with the ongoing data crisis

As with most AI-driven systems, advances in technology will eventually come to grips with the problem of data. Modeling the fringes of human biology requires much more sophisticated information than researchers can currently obtain.

“We still need real data from patients,” Aliper said. “The amount of data is very biased in the West, where it is produced. I think we need to do more at home, to have more accurate data, or real facts, so that our models are able to deal with it.”

Aliper said Insilico’s automated labs generate multi-species data from disease samples at scale, without human intervention, giving it its AI-powered diagnostic feature.

Zhu says the data AI needs already exists in the human body, built up over thousands of years of evolution. Only a fraction of the DNA represents the protein, while the rest acts as an instruction manual for how the genes function. This information has historically been difficult for humans to interpret but is increasingly being used by AI models, including recent efforts such as Google DeepMind’s AlphaGenome.

GenEditBio uses a similar approach in the lab, testing thousands of delivery nanoparticles in parallel instead of one at a time. The results, which Zhu calls “the gold of AI systems,” are used to train his models and, more importantly, to facilitate cooperation with external partners.

One of the next big things, according to Aliper, is creating digital twins of people for clinical trials, a process that he says is “still in progress.”

“We are on an island of about 50 drugs approved by the FDA Every year every year, and we should see growth,” said Aliper. “There is an increase in chronic diseases because we are aging as a population around the world…