Latent Labs Launches with $50M to Programmable Biology

Latent Labs Emerges with $50 Million to Revolutionize Protein Engineering

A recently established startup, spearheaded by a former scientist from Google DeepMind, has concluded its stealth phase and secured $50 million in funding.

Latent Labs is focused on developing AI foundation models with the goal of rendering biology programmable.

The Core of Biological Function: Proteins

Understanding the work conducted by DeepMind and similar organizations necessitates a firm grasp of the crucial role proteins play within human biology.

Proteins are the driving force behind all processes occurring within living cells, functioning as enzymes, hormones, and antibodies.

These complex molecules are constructed from approximately 20 different amino acids, linked together in chains.

These chains then fold into specific three-dimensional structures, and it is this shape that dictates the protein’s function.

AlphaFold and the Acceleration of Protein Structure Prediction

Historically, determining the structure of each protein was a protracted and demanding undertaking.

A significant advancement was achieved by DeepMind with AlphaFold, which integrated machine learning with authentic biological data.

This allowed for the prediction of the shapes of roughly 200 million protein structures.

Latent Labs: Designing Therapeutics from the Ground Up

With access to this wealth of structural data, scientists are now better equipped to decipher disease mechanisms, formulate novel drugs, and even engineer synthetic proteins for innovative applications.

Latent Labs aims to empower researchers to “computationally create” entirely new therapeutic molecules, marking a significant step towards programmable biology.

The company intends to collaborate with both biotech and pharmaceutical companies to generate and refine proteins for a variety of applications.

Latent Labs: Pioneering Protein Design

Simon Kohl, featured in the image above, initially built his career as a research scientist at DeepMind. He was a key member of the AlphaFold2 core team and subsequently co-led the protein design team, also establishing DeepMind’s wet lab facility at the Francis Crick Institute in London.

Concurrently, DeepMind fostered the creation of Isomorphic Labs, a sister company dedicated to applying DeepMind’s artificial intelligence research to revolutionize drug discovery. These parallel advancements led Kohl to believe the opportune moment had arrived to establish an independent, streamlined organization.

This new entity would concentrate on developing advanced models specifically for protein design. Consequently, Kohl left DeepMind in late 2022 to begin building Latent Labs, formally incorporating the business in London during mid-2023.

“My time at DeepMind was both rewarding and impactful, solidifying my conviction in the transformative potential of generative modeling within biology, particularly in protein design,” Kohl explained to TechCrunch this week. “However, with the launch of Isomorphic Labs and their AlphaFold2-based initiatives, they began pursuing numerous projects simultaneously. I recognized an opportunity to focus intensely on protein design.

Protein design represents an expansive field with substantial unexplored potential, and I believed a highly agile, dedicated team could effectively translate research into tangible results.”

Securing venture backing enabled the recruitment of a team of approximately 15 individuals. This included former colleagues from DeepMind, a senior engineer from Microsoft, and doctoral graduates from the University of Cambridge. Currently, Latent Labs operates across two locations.

One site is in London, serving as the hub for advanced model development, while the other is in San Francisco, housing both a wet lab and a computational protein design team.

“This dual-site approach allows us to validate our models through real-world experimentation and gather crucial feedback to ensure our models are evolving as intended,” Kohl stated.

While wet lab validation is a priority for confirming the accuracy of Latent’s technology, the long-term objective is to minimize the reliance on such facilities.

“Our core mission is to render biology programmable, effectively integrating it into the computational domain. This will progressively reduce the need for traditional, biological wet lab experiments,” Kohl clarified.

This ambition underscores a significant advantage of “making biology programmable” – fundamentally altering the drug discovery process, which currently depends on extensive experimentation and iterative cycles that can span years.

“It will empower us to create highly customized molecules without constant dependence on wet lab procedures – that is the ultimate vision,” Kohl elaborated. “Envision a future where researchers can formulate a hypothesis regarding a specific drug target for a disease, and our models can generate a protein drug with all desired characteristics, accessible with a simple interface.”

Latent Labs: A Novel Approach to Biological Research

Latent Labs distinguishes its business strategy by focusing on collaboration rather than internal drug development. The company intends to accelerate and minimize risk in the initial research and development phases through partnerships with external organizations.

According to Kohl, the greatest contribution Latent Labs can make is to empower other biopharmaceutical companies, biotechnology firms, and life science organizations. This will be achieved by providing direct access to their models or offering project-specific support for discovery initiatives.

Recent Funding and Investors

The company has secured $50 million in funding, comprising a prior $10 million seed investment and a new $40 million Series A round. This Series A round was jointly led by Radical Ventures, with partner Aaron Rosenberg – formerly of DeepMind’s strategy and operations team – taking a key role.

Sofinnova Partners, a French venture capital firm renowned for its life sciences investments, also co-led the funding round. Additional investors include Flying Fish, Isomer, 8VC, Kindred Capital, Pillar VC, and prominent individual investors.

These individual investors include Jeff Dean, Google’s chief scientist, Aidan Gomez, the founder of Cohere, and Mati Staniszewski, who founded ElevenLabs.

Allocation of Funds

A portion of the funding will be allocated to employee compensation, particularly for new machine learning specialists. However, a substantial amount will be dedicated to infrastructure costs.

“Significant computational resources are required for our work,” Kohl explained. “We are developing large models, which necessitates extensive GPU compute power.”

This funding will enable Latent Labs to expand its computational capacity, grow its teams, and establish the infrastructure needed to forge partnerships and achieve commercial success.

Competitive Landscape and Future Innovation

Several ventures, including Cradle and Bioptimus, are actively working to integrate computational methods with biological research. Kohl believes the field is still in its early stages, and the optimal strategies for understanding and engineering biological systems remain unclear.

“Groundbreaking work, such as AlphaFold and other generative models, has laid a foundation,” Kohl stated. “However, the field hasn’t yet settled on the most effective modeling techniques or viable business models.”

Latent Labs aims to capitalize on this opportunity and drive innovation within the field. The company believes it has the potential to significantly advance the understanding and design of biological systems.

• The company prioritizes partnerships over in-house drug development.
• Funding will support both personnel and crucial computational infrastructure.
• Latent Labs sees significant room for innovation in the intersection of biology and computation.