The amount of data in a single human genome is staggering. To understand how genes affect health, though, we need data from huge numbers of genomes. We also need to study the medical records of those whose genomes are analyzed. Proteomics and metabolomics make the task even more complex.
Proteomics
Proteomics is the large-scale study of proteomes. A proteome is a set of proteins produced in an organism, system, or biological context.
Metabolomics
Metabolomics is the large-scale study of small molecules , commonly known as metabolites, within cells, biofluids, tissues or organisms. Collectively, these small molecules and their interactions within a biological system are known as the metabolome.
----------
...Jim Mellon recently joined the race when the company he co-founded, Juvenescence Limited, announced a joint venture with Insilico Medicine. The joint venture is called Juvenescence AI. It will focus on developing compounds for the treatment of aging and age-related diseases...
I take Juvenescence AI seriously not just because Mellon has been right so often in the past. The computational side of the partnership, Insilico Medicine, may be the most driven company in AI today....
...Calico, Google’s medical venture...
...Many of Insilico’s researchers came to be there by winning online “hackathons”. These tests were designed to find the brightest AI experts and biologists, wherever they are. Also, most of the company’s R&D facilities are offshore. That means its costs are a fraction of its US counterparts’...
...The tools employed by Insilico Medicine are deep-learning neural networks that mimic human thought processes. These algorithms, run on supercomputers, analyze data too massive for human minds to comprehend.
Already, the company has identified dozens of compounds with potential to modify biological pathways involved in aging.
...Generative Adversarial Networks (GANs). GANs are neural networks capable of creation. GANs synthesize output based on specifications set by the programmer. You’ve probably heard about the GAN than can generate images of paintings in the style of specific artists based on user-provided descriptions.
...
One type of GAN is the generative Adversarial Autoencoders (AAE). AAEs can be trained to create content to satisfy the programmer’s rules. Imagine an AI that can be instructed to produce a set of molecules that prevent the deterioration of nerve cells in patients at risk for dementia.
AAEs can also predict adverse effects of a drug. AAEs can also evaluate the probability of a drug passing human clinical trials. Insilico Medicine has demonstrated the first proof of concept of AAEs in oncology.
AAEs have the potential to disrupt the entire drug development field. Once we know how a drug works, the program could be tasked with finding another compound that can bring about the same molecular changes.
This breakthrough has huge implications. For the first time, the potential exists to quickly reverse engineer drugs without violating their patents. While this may scare big drug companies, it’s good news for AI startups like Juvenescence AI and healthcare consumers.
Juvenescence AI will help improve Insilico Medicine’s AIs in return for the right to choose the best of its discoveries. According to Juvenescence CEO Jim Mellon and COO Alex Picket, among those candidates is at least one compound that appears to be a true anti-aging compound.
No comments:
Post a Comment