David Baker Serves As Scientific Advisor, Startup Releases World's Largest Protein Interaction Database, Receives 8 Rounds of Financing
Protein-Protein Interactions (PPI) are an important component of cellular life activities and play an indispensable role in regulating and maintaining cellular physiological functions (such as cell signal transduction, metabolic response and gene expression).
However, there are relatively few data in the PPl database, and the latest binding affinity data set (PDBbind+) only has 3,176 data.
Abihishaike Mahajan, a senior machine learning engineer at Dyno Therapeutics, pointed out in his article "Wet-lab innovations will lead the AI revolution in biology" that the revolutionary progress made by AlphaFold is extremely difficult to replicate. The most important reason is data.“We’re pretty much exhausting all the pre-existing data, and untrained protein structures and sequences are drying up.”
He also introduced that the characteristics of high-quality data in the biological field are that it has a complex potential distribution, is highly relevant to important physiological phenomena, and is suitable for large-scale collection. However, "the data types that meet the above three requirements have been exhausted."
Article URL:
https://www.owlposting.com/p/wet-lab-innovations-will-lead-the
Biotech startup A-Alpha Bio is trying to address this data shortage.Released AlphaSeq, the world's largest protein interaction database.The dataset contains more than 750 million measurements and is rapidly expanding at a rate of 3M-50M data points per month.
The past and present of the AlphaSeq dataset
The method used by AlphaSeq originated from a research paper published by David Baker's laboratory in 2017 titled "High-throughput characterization of protein-protein interactions by reprogramming yeast mating".The establishment of A-Alpha Bio is the transformation of this research result.
The paper introduces for the first time A-Alpha Bio's method for large-scale collection of protein-protein interaction data, demonstrating that SynAg can perform high-throughput quantitative characterization of protein-protein interaction networks at a library-to-library scale in a fully defined extracellular environment. This method is highly similar to the formally named AlphaSeq, so Mahajan referred to this method as AlphaSeq in his blog.
Paper address:
https://www.pnas.org/doi/10.1073/pnas.1705867114#sec-2
In 2022, A-Alpha Bio formally introduced the AlphaSeq dataset in the research paper "A dataset comprised of binding interactions for 104,972 antibodies against a SARS-CoV-2 peptide".This is a publicly available dataset consisting of 104,972 antibodies that interact with SARS-CoV-2 peptides.A quantitative measure containing antibody sequence, antigen sequence, and binding score is a benchmark for evaluating antibody specificity representation models in machine learning.
Paper address:
https://www.nature.com/articles/s41597-022-01779-4
In the latest study, A-Alpha Bio researchers used the AlphaSeq platform (which enables library-to-library screening of protein-protein interactions) to generate affinity-detuned cytokine therapeutic candidates and mouse-specific surrogates for fusion with antibodies targeting immune cell subsets. Using IFNA2 and IL-21 as examples, the study found that AlphaSeq can be used to identify hundreds of detuned cytokine variants with broad affinities for the IFNAR2 and IL21R receptors, respectively.
Paper address:
https://www.aalphabio.com/static/7a44b7b5f69a3e05eb02829a9278c470/ICIS_2023_poster_final.pdf
As an experimental platform,AlphaSeq can also quantitatively measure the binding affinity of millions of PPIs simultaneously.And the results can be obtained quickly, which perfectly meets the needs of large-scale expansion.
According to CTO Randolph Lopez, they currently perform about 30 AlphaSeq tests per month, and can obtain 100k~5M crossover points each time. This means that the AlphaSeq database is still expanding rapidly at a rate of 3M~50M per month.
A-Alpha Bio: A biotechnology company focused on PPI
A-Alpha Bio was founded in 2017 and is headquartered in Seattle, USA. It was founded by biologists from the Institute for Protein Design (IPO) and the Center for Synthetic Biology at the University of Washington. It is a biotechnology company that uses synthetic biology and machine learning to measure, predict and design protein-protein interactions.
The company built the experimental platform AlphaSeq and the computing platform AlphaBind. AlphaSeq is used to quantify and multiplex protein-protein interactions.Generate high-quality protein binding data at scale using synthetic biology approaches.This process builds on the connection between the strength of protein-protein interactions and yeast hybrids.
AlphaBind uses the accumulated data to train machine learning models that can predict protein sequence binding, test the models through experiments and quickly iterate and improve them, thereby accelerating drug discovery and designing better treatments.
A-Alpha Bio Technology Platform
Currently, AlphaSeq can quickly and quantitatively measure millions of protein-protein binding affinities simultaneously in a single experiment, and the trained AlphaBind can predict binding strength based on sequence.
A-Alpha Bio is based on the AlphaSeq + AlphaBind technology platform.We conduct business in multiple fields including antibody discovery and optimization, molecular glue target discovery, and machine learning model development.At the same time, the company has also developed an internal therapeutic pipeline. The self-developed pipeline mainly includes two parts: immune cytokine treatment pipeline and molecular glue target discovery pipeline, showing great development potential in the field of drug research and development.
Therefore, A-Alpha Bio has become the target of capital "hot pursuit".
A-Alpha Bio Internal Therapeutic Pipeline
According to public data, up to now,A-Alpha Bio has completed eight rounds of financing, with the highest financing reaching US$22.4 million.In May, A-Alpha Bio announced it had received $14.5 million in additional funding from the U.S. Department of Defense to further expand its partnership with Lawrence Livermore National Laboratory (LLNL), a federally funded research and development center.
Image source: official website press release
It is reported that the cooperation between A-Alpha Bio and LLNL began in 2022, and the two parties used synthetic biology and machine learning to accelerate the discovery of therapeutic antibodies against COVID-19 variants. At that time, A-Alpha Bio received $1 million in funding for this cooperation. By 2023, the two parties expanded the scale of cooperation to jointly develop drugs targeting multiple related pathogen families, and the company received another $2.4 million.
Today, the company has entered into partnerships with a number of companies and laboratories: such as conducting protein interaction research with Bristol Myers Squibb; collaborating with Gilead Sciences to explore HIV treatments; and collaborating with Kymera Therapeutics to discover and characterize new E3 ubiquitin ligases and rationally design molecular glues for high-value therapeutic targets, etc.
David Baker serves as scientific advisor
A-Alpha Bio was co-founded by two bioengineering PhDs from the University of Washington, David Younger and Randolph Lopez, with the goal of accelerating drug discovery by analyzing protein-protein interactions.
David Younger on the left and Randolph Lopez on the right
Image source: A-Alpha Bio official website
David Younger was a PhD student in David Baker's lab.During his doctoral studies, he focused on high-throughput protein interaction analysis and its application in bioengineering and clinical treatment. He developed innovative technologies based on the yeast two-hybrid system, especially characterizing protein-protein interactions by reprogramming yeast mating, which greatly improved the efficiency of protein-protein interaction screening and provided strong support for the development of AlphaSeq.
Randolph Lopez, another founder of A-Alpha Bio, mainly works in synthetic biology and computer science. He previously worked as a process development engineer at Illumina, responsible for improving quality, controlling efficiency and promoting products to large-scale production, providing a comprehensive foundation for the development and implementation of AlphaSeq + AlphaBind.
It is worth mentioning that A-Alpha Bio's scientific advisor is David Baker, a star scholar in the field of protein folding and a world-renowned biocomputing expert.The team he leads has achieved fruitful results in the fields of protein engineering and structural biology. They have developed protein structure prediction and design tools such as RoseTTAFold, ProteinMPNN, and RFdiffusion, and have won many awards including a nomination for the Nobel Prize in Physiology or Chemistry.
A-Alpha Bio Scientific Advisor (Image source: official website)
At the same time, A-Alpha Bio has also hired Chris Gibson, co-founder and CEO of Recursion Pharmaceuticals, Eric Klavins, professor of electrical engineering at the University of Washington, Tony Polverino, board member of Brainstorm Cell Therapeutics, and Lance Stewart, chief strategy and operations officer of the Institute for Protein Design (IPD) at the University of Washington School of Medicine, as scientific advisors to the company.
As of now, although A-Alpha Bio has not yet released the latest paper on AlphaSeq and detailed information on the AlphaBind model has yet to be disclosed, according to Mahajan's blog analysis, the application prospects of the AlphaSeq platform in the field of drug research and development are very broad.
As the AlphaSeq database continues to expand, we firmly believe that A-Alpha Bio will have more breakthroughs in drug research and development in the future to safeguard human life and health.
* A-Alpha Bio official website address:
References:
1. https://www.pdbbind-plus.org.cn/yuan
2. https://www.owlposting.com/p/wet-lab-innovations-will-lead-the
3. https://www.owlposting.com/p/creating-the-largest-protein-protein
4.https://www.vbdata.cn/companyDetail/9c03eefc878f2b4363d42bc9f84a97bc
5.https://bydrug.pharmcube.com/news/detail/dc7555da11321e7a5b5e3653fe46cf20
6.https://mp.weixin.qq.com/s/ve061fKiRi9Wbsvk8qvNag