Founded by Deepmind alumni, Latent Labs will be launched for $50 million to make biology programmable

A new startup established by the former Google DeepMind Scientists are finishing stealth with $50 million in funding.

Subject Lab We are building an AI foundation model to “make biology programmable” and will partner with biotechnology and pharmaceutical companies to generate and optimize proteins.

It is impossible to understand what Deepmind and its likeness do without first understanding the role proteins play in human biology. Proteins drive everything in living cells, from enzymes and hormones to antibodies. They are made up of about 20 different amino acids, and are joined by strings that fold to create a 3D structure, whose shape determines the function of the protein.

However, understanding the shape of each protein has historically been a very slow and labor-intensive process. That was a big breakthrough DeepMind achieved with Alphafold: It meshed real biological data and machine learning to predict the shape of approximately 200 million protein structures.

Armed with such data, scientists can better understand diseases, design new drugs, and more Creates synthetic proteins For a completely new use case. So, potential labs enter the fight with the ambition to allow researchers to “create” new therapeutic molecules from scratch.

Potential potential

Simon Cole (Photo above) Beginning as a research scientist at DeepMind, working with the Core alphafold2 With the team before they co-starred with the protein design team. Set up a DeepMind wet lab At the Francis Crick Institute in London. Around this time, Deep Mind also created a sister company. In the form of a lab,This focuses on applying deepmind’s AI research to transform drug discovery.

This was a combination of these developments that convinced Cole that it was right to do it alone in a more learmer outfit that focused specifically on building frontier (i.e. cutting edge) models for protein design. So, at the tail end of 2022, Cole set out a deep attitude to lay the foundations for a potential lab and incorporated the business in London in mid-2023.

“I had a fantastic and impactful time (at deepmind), and I was particularly convinced of the impact that production modeling has on biology and protein design,” Cole told TechCrunch in an interview this week. “At the same time, I saw it at the launch of the Essay Lab. Planning based on Alphafold2that they had started doing a lot at once. This opportunity felt like it was really going to go in a laser-focused way about protein design. The design of protein itself is a very vast field, with so many unexplored white spaces that I thought a truly agile and focused outfit could translate that influence. ”

It employed around 15 employees to translate the impact as a venture-backed startup. Two of them were senior engineers at Microsoft and received their PhD from Cambridge University. Today, Latent personnel are divided into two sites. One is in London, the frontier model magic occurs, the other is in San Francisco, on its own site Wet Lab Computational protein design team.

“This allows us to test our models in the real world and get the feedback we need to understand if the models are going the way we want them to,” Cole said.

Wet Labs are on a very close agenda in terms of examining Latent’s technology predictions, but the ultimate goal is to deny the need for wet Labs.

“Our mission is to make biology programmable, and we actually bring biology into the field of computation, where our reliance on biological wet lab experiments decreases over time. “Cole said.

This highlights one of the key benefits of “making biology programmable.” It now reverses the drug discovery process that relies on countless experiments and iterations that could take years.

“This allows us to really create custom molecules without resorting to wet labs. At least that’s the vision,” Cole continued. “Imagine a world where someone is hypothesizing about drug targets chasing a particular disease. Our model uses a ‘push button’ method to create a protein drug with all the desired baked properties. You can make it. . ”

Biology Business

When it comes to business models, potential labs do not consider themselves “asset-centric.” This means that we will not develop our own treatment candidates within the company. Instead, we would like to work with third-party partners to speed up the previous R&D stage and extend the risk.

“Our biggest impact as a company is enabling other biopharma, biotechnology and life science companies. We can either have direct access to the model or support discovery programs through project-based partnerships,” Cole said. said.

The company’s $50 million cash injection includes a previously unannounced $10 million seed tranche. Aaron Rosenbergpreviously head of strategy and operations at DeepMind.

Another co-lead investor is Sofinnova Partners, a French VC company that has recorded a long trajectory in the life sciences space. Other participants in the round include Flying Fish, Isomer, 8VC, Kindred Capital, Pillar VC and Google’s chief scientist Jeff Dean, Core founder Aidan Gomez, and Eleven Love founder Matistanishevski. includes famous angels such as.

Cash chunks, including new machine learning employment wages, will be directed towards pay, but will require a substantial amount of money to cover the infrastructure.

“Computing is a huge cost for us too. We’re building a pretty big model that I think is fair to say, and that requires a lot of GPU calculations,” Cole said. Ta. “This funding really has made it all double down, and we’re starting to scale our models, scale our teams, build bandwidth and these partnerships and the commercial traction we’re looking for. I’ll win it.”

Aside from DeepMind, there are several venture-backed startups and scale-ups. Cradles etc. and Bioptimus. Cole believes we are still in the early stages on his part.

“There are very interesting seeds planted (for example) with alphafolds and other early generative models from other groups,” Cole said. “But this area hasn’t converged in terms of what the best model approach is, or which business models work here. I think they have the ability to innovate really.”