Founded by DeepMind alumnus, Latent Labs launches with $50M to make biology programmable

A new startup founded by a former Google DeepMind scientist leaves Stealth with $ 50 million in financing.

Latent laboratories Builds AI Foundation models to ‘make biology programmable’ and it is planning to collaborate with biotech and pharmaceutical companies to generate and optimize proteins.

It is impossible to understand what DeepMind and his peers do without first understanding the role that proteins play in human biology. Proteins drive everything in living cells, from enzymes and hormones to antibodies. They consist of approximately 20 different amino acids, which connect to each other in strings that fold to create a 3D structure, the shape of which determines how the protein functions.

But sorting out the shape of each protein was historically a very slow, labor -intensive process. That was the big breakthrough that DeepMind reached with Alphafold: it took machine learning with real biological data to predict the form of around 200 million protein structures.

Armed with such data, scientists can better understand diseases, design new medicines and even Make synthetic proteins For completely new use cases. That is where latent laboratories enter the battle with his ambition to enable researchers to ‘make’ new therapeutic molecules completely.

Latent

Simon Kohl (shown above) started as a research scientist at DeepMind, together with the Core Alpafold2 team before he led the protein design team and Setting up DeepMind’s Wet Laboratory At Francis Crick Institute in London. Around this time, DeepMind also produced a sister company in the form of Isomorfe Laboratories, which is aimed at applying DeepMind’s AI research to transform the discovery of drugs.

It was a combination of these developments that convinced Kohl that the time was good to do it alone with a slimmer outfit that was specifically aimed at building frontier (ie advanced) models for protein design. So at the end of 2022, Kohl DeepMind left to lay the foundation for latent laboratories and to include the company in London in mid -2023.

“I had a fantastic and impactful time [at DeepMind]And was convinced of the impact that generative modeling would have in particular in biology and protein design, ”Kohl told WAN in an interview this week. “At the same time, I saw that with the launch of Isomorfe Laboratories and their plans based on Alpafold2, that they started many things at the same time. I felt that the chance was to be about protein design in a laser -oriented way. Protein design is in itself such a vast field and has so much unexplored white space that I thought a truly agile, targeted outfit could translate that impact. “

Translating that impact as a startup supported by the company included hiring around 15 employees, of whom two of DeepMind, a senior engineer from Microsoft and PhD students were from the University of Cambridge. Today the workforce of Latent is spread over two locations – one in London, where the frontier model magic happens, and another in San Francisco, with its own wet laboratory and computational protein design team.

“This enables us to test our models in the real world and get the feedback that we have to understand or our models are progressing the way we want,” said Kohl.

Although wet laboratories are the agenda in the short term in terms of validating the predictions of Latent’s technology, the ultimate goal is to destroy the need for wet laboratories.

“Our mission is to make biology programmable, so that biology is really brought into the computational empire, where dependence on biological, wet lab experiments will be reduced over time,” Kohl said.

This emphasizes one of the most important benefits for “making biology programmable”-increasing a drug description process that currently depends on countless experiments and iteration that can last for years.

“It enables us to make really adapted molecules without trusting the wet lab – at least, that’s the vision,” Kohl continued. “Imagine a world where someone comes with a hypothesis about which medicine objective you should do for a certain disease, and our models can make a protein medicine on a ‘push button’ that is accompanied by all the desired properties that are built in. “

The company of biology

In terms of business model, latent laboratories do not see themselves as ‘asset-centric’-what means that it will not develop its own therapeutic candidates in-house. Instead, working with third-party partners wants to accelerate and risk the earlier R&D phases.

“We believe that the biggest impact we can have as a company is by making other biopharma, biotechs and life science companies possible by giving them direct access to our models, or supporting their discovery programs through project-based partnerships , “said Kohl.

The cash injection of $ 50 million from the company includes a previously unannounced seed tranche of $ 10 million and a fresh series of $ 40 million per round led by radical companies-in name partner Aaron RosenMountainThat previously was head of strategy and operations at DeepMind.

The other Co-Lead Investor is Sofinnova Partners, a French VC company with a long track record in the space of Life Sciences. Other participants in De Ronde are Flying Fish, Isomer, 8VC, Kindred Capital, Pillar VC and remarkable angels such as the chief scientist Jeff Dean of Google, founder Aidan Gomez and Elflabs founder Mati Staniszewski.

Although part of the money goes to salaries, including those of new Machine Learning recruitment, a considerable amount will be needed to cover the infrastructure.

“Calculation is also a major costs for us – we build quite large models that I think is fair to say, and that requires a lot of GPU calculations,” said Kohl. “This financing really sets us to double everything on everything – acquires Compute to keep scaling up our model, to scale the teams and also start building the bandwidth and the ability to have these partnerships and the commercial traction we have Search now. “

Apart from DeepMind, there are several startups supported by Venture and scaling up the worlds of calculation and biology closer together, such as Cradle and BiopTimus. Kohl, for his part, thinks that we are still at a sufficiently early stage, where we still do not fully know what the best approach will be in terms of decoding and designing organic systems.

“Some very interesting seeds have been planted, [for example] With Alpafold and some other early generative models from other groups, “said Kohl. “But this field is not combined in terms of what the best model approach is, or in terms of which business model will work here. I think we have the ability to really innovate. ‘