Insilico Medicine: The World’s First AI Driven Mechanism for Idiopathic Pulmonary Fibrosis Drugs

0
110

Insilico Medicine has made breakthroughs in artificial intelligence and new drug development—for the first time, it combines biology and chemical generation to discover a new clinical candidate drug for the treatment of idiopathic pulmonary fibrosis (IPF) with a new mechanism, and successfully passed multiple Sub-human cell and animal model experimental verification. IPF involves multiple diseases and affects multiple organs (lung, liver, and kidney). The emergence of this new drug is expected to solve the widespread unmet medical needs that affect thousands of people around the world.

The cause of IPF is still unknown, and the pathogenesis of IPF is still unknown in the medical community. The disease is mostly sporadic. The average survival time of patients from the onset of symptoms to death is no more than 5 years.

Extensive pulmonary fibrosis easily complicates lung cancer, and pulmonary hypertension can also occur in the late stage. The drugs currently used to treat IPF have been used clinically for more than 30 years, and only have an effect on 10% to 30% of patients. Patients rely on oxygen therapy to improve their quality of life in the late stage of the disease, but the situation is not optimistic.

Dr. Alex Zhavoronkov, founder and CEO of Insilico Medicine, said: “Associating the right drug target with the right disease is the biggest challenge in drug development.” “As today we achieve the first artificial intelligence discovery and scientifically validated clinical In the former drug candidate (PCC) milestone, Insilico has overcome another major obstacle in drug discovery and broke through another bottleneck in the traditional drug discovery process, which took very little cost and time.”

AI rewrites the history of drug discovery

From target discovery to the invention of pre-clinical drug candidates, Insilico achieved target discovery, molecular generation, and traditional experimental verification in less than 18 months, and confirmed the efficacy and safety of IPF in animals. The total cost is approximately The total cost of research on the efficacy of other fibrotic diseases is about US$1.8 million, and no more than 80 small molecule compounds have been synthesized and tested.

Traditional drug discovery first involves testing and screening tens of thousands of small molecules, and then further synthesizing and testing hundreds of molecules in order to obtain a few drug candidates suitable for preclinical research, of which only about 1/10 of the drug candidates can finally Through clinical trials in human patients. The entire process is slow and costly. It takes an average of 10 years and costs hundreds of millions of dollars.

Another obstacle that further hinders the introduction of new drugs to the market is the large number of R&D steps involved in the entire R&D process—each stage costs hundreds to tens of millions of dollars—often by different companies or different business units in the drug R&D industry. .

Dr. Zhavoronkov stated: How AI discovers new mechanisms of idiopathic pulmonary fibrosis drugs

Insilico Medicine started researching 20 new potential targets related to fibrosis discovered through artificial intelligence, and gradually narrowed the scope of indications to a new target specifically for IPF.

After the target was determined, Insilico designed a new set of compounds through the artificial intelligence chemical generation system to selectively inhibit this new target. These molecules must have good selectivity, bioavailability, metabolic stability, oral administration properties, safety, and multiple high-quality properties unique to drugs. These molecules were originally generated by the structure-based molecular design algorithm in Chemistry42, the company’s generative chemistry artificial intelligence system, and showed their effectiveness in cell experiments and animal model experiments.

These experimental data are then fed back to the artificial intelligence system, and the artificial intelligence redesigns a new batch of compounds to optimize the activity and druggability, and verify again.

After several rounds of design-synthesis-assessment-optimization-redesign cycles, preclinical candidate compounds have been identified. Insilico’s preclinical candidate compounds have passed the strict evaluation of the company’s internal and external fibrotic disease experts, and have entered the preclinical research stage.

In addition, the company also uses artificial intelligence to predict the success rate of the phase II clinical trial of this new IPF target and new molecule. Insilico is currently conducting an IND application experiment, with the goal of conducting clinical studies in early 2022.

Insilico welcomes and looks forward to working with pharmaceutical companies to jointly develop drugs after Phase II.

Although the hot topics surrounding the research and development of new drugs usually focus on when a new target is discovered or when a new drug enters clinical trials, the areas that are most suitable for innovation and have the greatest business impact are from target discovery to clinical development.

Insilico makes history

In 2019, Insilico made history. It invented and launched a new artificial intelligence system for drug discovery that can create brand-new molecules from start to finish in 21 days at a cost of only about $150,000. Since the failure rate of target discovery is about 95%, Insilico solved one of the biggest bottlenecks in drug discovery in the industry. Insilico’s artificial intelligence software is driven by generative chemistry using modern artificial intelligence technology, which can quickly generate new molecular structures with specific properties.

As the first company to explore the use of generative confrontation network (GAN) and generative reinforcement learning (RL) artificial intelligence technology for drug discovery, the success of Insilico’s artificial intelligence software is the first to show the industry the first successful discovery and generation of new clinical candidate compounds Scientifically verified.

Dr. Zhavoronkov said: “The peak of the deep learning revolution can be traced back to 2014, when a generative adversarial network appeared, and the deep learning system began to surpass humans in the field of image recognition. In the same year, the company was established. In 2016, we verified through experiments that the deep learning system can be Identify new biological targets in scientific data. From 2017 to 2019, we have continued to prove that generative artificial intelligence can invent and design new molecules that are active in human cells and animals.

But there is still a big problem-can artificial intelligence design a new molecule for a new target that has no known inhibitors and has not been verified in the disease? Now, we have successfully combined biology and chemistry, and have been nominated for a preclinical drug candidate that can act on a new target. The purpose is to use it in human clinical trials. This is an urgent need to solve and an order of magnitude more Complex and riskier problems.

As far as I know, this is the first case of artificial intelligence successfully discovering a new target and designing a new preclinical drug candidate that can act on disease indications in a large population. This is an important milestone for us. Our final “Moon Landing Program” is to solve the problem of human aging, which requires us to have more and more reliable artificial intelligence technologies to help us understand and regulate human biology in other chronic diseases. ”

In addition, Insilico will receive huge financial support for drug discovery and development on a variety of new drug targets. The company has used the independently developed Pharma.AI software to provide target discovery and generative chemical system services and support for pharmaceutical and biotechnology companies. The PandaOmics target discovery AI system can be provided as a software service. The Chemistry42 small molecule generation chemistry platform has been installed and deployed on the site of pharmaceutical users in September 2020. So far, the world’s most advanced pharmaceutical companies have begun to adopt our Chemistry42 molecule generation and design platform, and PandaOmics has been adopted in the drug target discovery departments of many famous academic institutions and pharmaceutical companies.

Insilico also announced that the company will continue to grow its scientific research team. It has established a team of more than 20 senior drug developers in Shanghai, led by Chief Scientific Officer (CSO) Dr. Feng Ren, who joined Insilico in February this year. Prior to this, he successively served as Senior Vice President of the Biology and Chemistry Department of Medicilon Biopharmaceuticals, and Director of Chemistry of GSK GlaxoSmithKline. The team is responsible for advancing new drug projects discovered by artificial intelligence to clinical trials and creating a broad portfolio of preclinical/clinical drug products.

About Insilico Medicine

The software developed by Insilico Medicine uses generative models (GAN), reinforcement learning (RL) and other modern machine learning techniques to generate new molecular structures with specific properties. Insilico Medicine has also developed software for generating molecules, identifying targets, and predicting clinical trial results. The company integrates two business models: through the self-developed Pharma.AI platform (www.insilico.com/platform/), it provides artificial intelligence-driven drug discovery services and software, and independently develops preclinical and clinical projects. The pre-clinical project is realized through its own platform to find new drug targets and new molecules. Since its establishment in 2014, Insilico Medicine has raised more than US$52 million in funds and won multiple industry awards. Insilico Medicine has also published more than 100 peer-reviewed papers and has applied for more than 25 patents.

LEAVE A REPLY

Please enter your comment!
Please enter your name here