#Xeueka

[govindhtech]

Xeureka Aims To Accelerate Durg Research And Development

How Xeureka is Revolutionizing Drug Research and Development

Using strong models developed on data sets that can be securely shared owing to the NVIDIA AI platform, a Mitsui & Co. subsidiary seeks to speed up drug development.

Mitsui & Co. was founded 77 years ago and continues to thrive by creating ecosystems and enterprises using cutting-edge technologies like confidential computing and generative AI.

At the 16-division conglomerate centered in Tokyo, digital change takes numerous shapes. It might be a geospatial analytic platform in one instance or an autonomous trucking service in another. In fact, Mitsui works with a partner at the forefront of quantum computing.

A new subsidiary, Xeureka, wants to speed up the healthcare industry’s billion-dollar, ten-year drug development process research and development.

Katsuya Ito, Mitsui’s digital transformation project manager, stated, “It use new digital technology like AI and confidential computing to create businesses.” The bulk of they work is carried out in partnership with IT firms, in this example Fortanix, a security software business located in San Francisco, and NVIDIA.

In Pursuit of Big Data

Despite being only three years old, Xeureka has already finished a proof of concept that tackles one of the main issues with drug discovery gathering sufficient data.

Strong AI models developed using datasets greater than those available to most pharmaceutical corporations are necessary to speed up drug development. Since data often includes confidential patient information and pharma company-proprietary chemical formulations, sharing between firms has been unimaginable until recently.

Confidential computing is the method of processing data in a secure area of a GPU or CPU that serves as a “black box” for the most critical information of a company.

Banks, government organizations, and even advertising use the technology, which is supported by a group of some of the biggest businesses in the world, to guarantee that their data is always kept private.

A Proof of Concept for Privacy

Xeureka established two fictitious businesses, each with a thousand drug candidates, to confirm that private computing would enable its clients to securely exchange data. An AI model was trained using the datasets of each firm independently to forecast the toxicity levels of the compounds. A bigger but comparable AI model was then trained using the combined data.

Using security management software from Fortanix, one of the first firms to provide secret computing, Xeureka tested on NVIDIA H100 Tensor Core GPUs.

With hardware-based engines that guarantee and verify that private workloads are safeguarded while being used on the GPU, the NVIDIA H100 GPUs enable a trusted execution environment without sacrificing speed. Data exchange, encryption keys, and workflow management are all handled by the Fortanix program.

Up to 74% Higher Accuracy

The results were striking. With the pooled datasets, the larger model predicted 65-74% more accurately.

According to Ito, the models developed using data from a single business had bias and instability that were absent from the bigger model.

In a Fortanix press release, Xeureka CTO Hiroki Makiguchi stated, “Confidential computing from NVIDIA and Fortanix basically alleviates privacy and security concerns while improving model accuracy, which will prove to be a win-win situation for the entire industry.”

An AI Supercomputing Ecosystem

Now, Xeureka is working with the community behind Tokyo-1, its GPU-accelerated AI supercomputer, to investigate wide-ranging uses of this technology in drug discovery research. Tokyo-1, which was announced in February, is to improve pharmaceutical businesses’ efficiency both in Japan and abroad.

Partnerships to screen ligand-base pairs, predict protein structures, and speed up molecular dynamics simulations with reliable services might be examples of early efforts.Tokyo-1 users may use NVIDIA BioNeMo, drug discovery microservices and framework to apply large language model for chemical, protein, DNA, and RNA data.

It is a component of Mitsui’s larger strategic expansion goal to create healthcare software and services, including powering Japan’s $100 billion pharmaceutical business, which is the third biggest in the world after the United States and China.

Xeueka will employ AI to model intricate chemical processes, anticipate how beneficial chemicals will connect with proteins, and rapidly screen billions of medication candidates.

Read more on Govindhtech.com