AI for Drug Discovery from Scratch

An Artificial Intelligence (AI) system, dubbed ReLeaSE (see below), developed by University of North Carolina scientists, has demonstrated AI-based drug discovery from scratch. In a preliminary study, the scientists used ReLeaSE to generate molecules with properties that they specified, such as desired bioactivity and safety profiles. The team used the ReLeaSE method to design molecules with customized physical properties, such as melting point and solubility in water, and to design new compounds with inhibitory activity against an enzyme that is associated with leukemia.

Commenting on how AI is transforming drug creation, Alexander Tropsha, one of the creators of ReLeaSE, said, "The ability of the algorithm to design new, and therefore immediately patentable, chemical entities with specific biological activities and optimal safety profiles should be highly attractive to an industry that is constantly searching for new approaches to shorten the time it takes to bring a new drug candidate to clinical trials."

AI system trains itself to design new drugs. Scientists at UNC Chapel Hill have created an Artificial Intelligence (AI) system that can teach itself to design new drug molecules from scratch. The system, dubbed ReLeaSE (Reinforcement Learning for Structural Evolution), knows the chemical structures of about 1.7 million known biologically active molecules, and can train itself to propose molecules that are likely to be useful as new medicines. A proof-of-concept study on drug discovery with AI is published in Science Advances. The researchers are persuaded that AI drug research has the potential to dramatically accelerate the design of new drug candidates.

Stopping Alzheimer’s disease before symptoms appear. Biologists at University of Virginia have gained new understanding of how Alzheimer's disease begins, and how it might be halted using a current medication. A study published in Alzheimer's & Dementia suggests that an FDA-approved drug, memantine, currently used only for alleviating the symptoms of moderate-to-severe Alzheimer's disease, might be used to prevent or slow the progression of the disease if used before symptoms appear.

Turning off protein could improve cancer immunotherapy. Researchers at Johns Hopkins Medicine have discovered that inhibiting a previously known protein could enhance the effectiveness of immunotherapy treatments. The scientists studied laboratory mice genetically engineered to lack the Yes-associated protein (YAP) in several immune T-cell populations, including regulatory T-cells known as Tregs. A study published in Cancer Discovery shows that YAP plays a role in the suppression of anti-tumor immunity by Tregs. By turning off YAP's abilities, tumor killing with less restrained immune cells is possible.

DNA repair pathway could boost the efficiency of CRISPR-Cas9 gene editing. Scientists at UC Berkeley have discovered that a well-known DNA repair pathway, the Fanconi anemia pathway, plays a key role in repairing double-strand DNA breaks created by CRISPR-Cas9 gene editing. A research paper published in Nature Genetics indicates that the the Fanconi anemia pathway acts as a “traffic cop” to steer repair to simple end-joining or to patching the cut with new, single-strand DNA. The discovery could permit boosting the efficiency with which cells insert new DNA into the genome -- to replace a harmful mutation with the correct DNA sequence, for example -- and generally tweak CRISPR-Cas9 editing to get the desired outcome.

Switching on gene associated with hair loss could boost cancer immunotherapy. Researchers at Columbia University have found that a gene associated with an autoimmune form of hair loss may be activated to boost cancer immunotherapy. A study published in Cell Systems shows that a gene that recruits immune T cells in alopecia areata -- a condition in which immune cells attack and destroy hair cells -- is turned off in various types of cancer, protecting them from the immune system. Turning that gene back on can make those cancers vulnerable to the immune response.

Magnetic nanoparticles deliver chemotherapy drug across the blood-brain barrier. Researchers at UI Chicago have demonstrated that magnetic nanoparticles can be used to ferry chemotherapy drugs into the spinal cord to treat hard-to-reach spinal tumors in an animal model. The new delivery system, based on tiny, metallic magnets bound to particles of chemotherapy drug doxorubicin, described in a research paper published in Scientific Reports, represents a novel way to target chemotherapy drugs to spinal cancer cells, which are hard to reach because the drugs must cross the blood-brain barrier.