Deep Genomics Unveils DG12P1: Industry’s First AI-Discovered Therapeutic Candidate Spanning Target Identification to Compound Design

When it comes to using artificial intelligence (AI) to assist in new drug development, we may not be unfamiliar with this concept. However, AI is typically used to assist in individual steps of new drug development, such as building disease models to help discover new targets; or based on specific targets, accelerating the screening of hit compounds and lead compounds; or re-examining existing drugs to explore their potential for treating new indications.

Deep Genomics Inc. in Canada announced today that its AI-driven drug discovery platform guides the entire process from innovative therapeutic target identification to lead compound screening, completing the de novo generation of an innovative drug candidate within 18 months! The next step for this AI-discovered candidate is to conduct studies supporting an Investigational New Drug (IND) application. The press release indicates that this is the industry’s first therapeutic candidate discovered by AI!

Deep Genomics Inc. was founded in 2014 by Dr. Brendan Frey, a Canadian scholar and expert in machine learning and genomic biology. The company’s drug discovery platform comprises more than 20 machine learning systems specialized in analyzing the impact of genetic mutations on protein function and expression. Although the human genome contains millions of genetic variants, their associations with diseases are often difficult to decipher. According to Dr. Frey, the AI system can “scan 200,000 disease-causing genetic mutations within two hours and automatically select promising drug targets.” Rather than predefining specific disease areas, Deep Genomics allows its AI system to independently identify targets for further development. After screening over 2,400 diseases and more than 120,000 genetic mutations, and analyzing potential targets based on approximately 150 criteria, the AI system determined that a genetic mutation associated with Wilson Disease warranted further investigation.

Wilson’s disease is a rare genetic disorder in which patients lose the ability to excrete copper, leading to excessive accumulation of copper in the body. If left untreated, this condition can be fatal. Current treatments aim to prevent the absorption of copper from food or to enhance its excretion through urine. However, over the past two decades, although scientists have identified certain gene mutations associated with Wilson’s disease, they have been unable to clarify the pathogenic mechanisms underlying these mutations. Consequently, targeted therapies capable of restoring the body’s ability to excrete copper have not yet been developed.

Deep Genomics Inc.’s AI system has achieved a breakthrough in this area. By analyzing genetic mutations, it identified that the variant causing the Met645Arg mutation in the ATP7B protein leads to exon skipping during the splicing of the RNA encoding this protein. ATP7B is a copper-binding protein expressed in hepatocytes and plays a critical role in transporting excess copper into bile. Previously, although scientists had identified the Met645Arg mutation, experimental studies indicated that the protein carrying this mutation did not exhibit functional loss. However, the AI system revealed that the exon skipping caused by this mutation results in premature termination of ATP7B protein synthesis, thereby reducing ATP7B levels and impairing copper excretion. This discovery identifies a gene target with a clearly defined pathogenic mechanism for the treatment of Wilson’s disease. The study was published on bioRxiv.

Subsequently, the AI platform screened thousands of compounds to identify 12 lead candidates. Researchers validated the properties of these compounds in the laboratory and conducted further tolerance and pharmacokinetic studies. Ultimately, they selected a compound named DG12P1 for preclinical studies to support an Investigational New Drug (IND) application. This candidate is an oligonucleotide drug that functions similarly to Spinraza, approved by Biogen for the treatment of spinal muscular atrophy (SMA), by binding to precursor RNA and modulating RNA splicing. Throughout the early drug discovery process, the AI system provided guidance at every step, including specifying the chemical parameters for the oligonucleotides to be synthesized.

▲Deep Genomics’ interdisciplinary team (Image source: Deep Genomics official website)

Deep Genomics Inc.’s 40-person team plans to submit an Investigational New Drug (IND) application by the end of next year. In addition to this candidate therapy, the company is conducting R&D in metabolic diseases, ophthalmology, and neurodegenerative diseases. Dr. Frey stated, “We will launch two drug candidates this year, expect at least a doubling of the number of candidates next year, and another at least twofold increase the following year.”

In the field of new drug development, it is often said that the easy targets—the “low-hanging fruit”—have already been picked. Future drug discovery will require reaching for fruits high up in the tree, obscured by dense branches and leaves. Dr. Frey believes that AI will serve as the ladder enabling us to climb higher into the tree and harvest these fruits. Whether his prediction will come true remains to be seen.

References:

[1] Deep Genomics Nominates Industry’s First AI-Discovered Therapeutic Candidate. Retrieved September 25, 2019, from https://www.deepgenomics.com/updates/deep-genomics-nominates-industrys-first-ai-discovered-therapeutic-candidate/

[2] Merico et al., (2019). ATP7B Variant c.1934T>G p.Met645Arg Causes Wilson Disease by Promoting Exon 6 Skipping. BioRxiv. doi: https://doi.org/10.1101/693572.

[3] Canadian star researcher Brendan Frey brandishes Deep Genomics' first AI-discovered drug. Retrieved September 25, 2019, from https://endpts.com/canadian-star-researcher-brendan-frey-brandishes-deep-genomics-first-ai-discovered-drug/

[4] Deep Genomics reveals its program: The first AI-discovered drug candidate. Retrieved September 25, 2019, from https://www.fiercebiotech.com/biotech/deep-genomics-reveals-its-program-first-ai-discovered-drug-candidate

▽Follow [WuXi AppTecDe】WeChat Official Account