How AI is helping solve the labor issue in treating rare diseases

Modern biotech has the tools to edit genes and design drugs, yet thousands of rare diseases remain untreated. According to executives from Insilico Medicine and GenEditBio, the missing ingredient for years has been finding enough smart people to continue the work. AI, they say, is becoming the force multiplier that lets scientists take on problems the industry has long left untouched. 

Speaking this week at Web Summit Qatar, Insilico’s president, Alex Aliper, laid out his company’s aim to develop “pharmaceutical superintelligence.” Insilico recently launched its “MMAI Gym” that aims to train generalist large language models, like ChatGPT and Gemini, to perform as well as specialist models.

The goal is to build a multimodal, multitask model that, Aliper says, can solve many different drug discovery tasks simultaneously with superhuman accuracy.

“We really need this technology to increase the productivity of our pharmaceutical industry and tackle the shortage of labor and talent in that space, because there are still thousands of diseases without a cure, without any treatment options, and there are thousands of rare disorders which are neglected,” Aliper said in an interview with TechCrunch. “So we need more intelligent systems to tackle that problem.”

Insilico’s platform ingests biological, chemical, and clinical data to generate hypotheses about disease targets and candidate molecules. By automating steps that once required legions of chemists and biologists, Insilico says it can sift through vast design spaces, nominate high-quality therapeutic candidates, and even repurpose existing drugs — all at dramatically reduced cost and time.

For example, the company recently used its AI models to identify whether existing drugs could be repurposed to treat ALS, a rare neurological disorder. 

But the labor bottleneck doesn’t end at drug discovery. Even when AI can identify promising targets or therapies, many diseases require interventions at a more fundamental biological level. 

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

“We have developed a proprietary ePDV, or engineered protein delivery vehicle, and it’s a virus-like particle,” GenEditBio’s co-founder and CEO, Tian Zhu, told TechCrunch. “We learn from nature and use AI machine learning methods to mine natural resources and find which kinds of viruses have an affinity to certain types of tissues.”

The “natural resources” Zhu is referring to is GenEditBio’s massive library of thousands of unique, nonviral, nonlipid polymer nanoparticles — essentially delivery vehicles designed to safely transport gene-editing tools into specific cells.

The company says its NanoGalaxy platform uses AI to analyze data and identify how chemical structures correlate with specific tissue targets (like the eye, liver, or nervous system). The AI then predicts which tweaks to a delivery vehicle’s chemistry will help it carry a payload without triggering an immune response. 

GenEditBio tests its ePDVs in vivo in wet labs, and the results are fed back into the AI to refine its predictive accuracy for the next round. 

Efficient, tissue-specific delivery is a prerequisite for in vivo gene editing, says Zhu. She argues that her company’s approach reduces the cost of goods and standardizes a process that has historically been difficult to scale. 

“It’s like getting an off-the-shelf drug [that works] for multiple patients, which makes the drugs more affordable and accessible to patients globally,” Zhu said. 

Her company recently received FDA approval to begin trials of CRISPR therapy for corneal dystrophy.

Combating the persistent data problem

As with many AI-driven systems, progress in biotech ultimately runs up against a data problem. Modeling the edge cases of human biology requires far more high-quality data than researchers currently can get. 

“We still need more ground truth data coming from patients,” Aliper said. “The corpus of data is heavily biased over the Western world, where it is generated. I think we need to have more efforts locally, to have a more balanced set of original data, or ground truth data, so that our models will also be more capable of dealing with it.”

Aliper said Insilico’s automated labs generate multi-layer biological data from disease samples at scale, without human intervention, which it then feeds into its AI-driven discovery platform. 

Zhu says the data AI needs already exists in the human body, shaped by thousands of years of evolution. Only a small fraction of DNA directly “codes” for proteins, while the rest acts more like an instruction manual for how genes behave. That information has historically been difficult for humans to interpret but is increasingly accessible to AI models, including recent efforts like Google DeepMind’s AlphaGenome. 

GenEditBio applies a similar approach in the lab, testing thousands of delivery nanoparticles in parallel rather than one at a time. The resulting datasets, which Zhu calls “gold for AI systems,” are used to train its models and, increasingly, to support collaborations with outside partners. 

One of the next big efforts, according to Aliper, will be building digital twins of humans to run virtual clinical trials, a process that he says is “still in nascence.”

“We’re in a plateau of around 50 drugs approved by the FDA every year annually, and we need to see growth,” Aliper said. “There is a rise in chronic disorders because we are aging as a global population … My hope is in 10 to 20 years, we will have more therapeutic options for the personalized treatment of patients.”