Recursion Unveils OpenPhenom-S/16 on Google Cloud
Recursion, a prominent player in the clinical stage TechBio sector, has introduced an exciting development in the realm of drug discovery with the launch of OpenPhenom-S/16. This foundation model becomes publicly available on Google Cloud’s Vertex AI Model Garden, allowing life science companies to leverage advanced microscopy data to enhance their drug treatment assessments, analyze biological cultures, investigate tissue samples, and ultimately accelerate the pace of drug discovery.
Overview of OpenPhenom-S/16
According to Ben Mabey, Recursion’s Chief Technology Officer, this groundbreaking model is poised to establish a new gold standard in the industry. OpenPhenom-S/16 has been meticulously trained using a masked-autoencoder-based self-supervised learning approach, utilizing a colossal dataset comprising over three million microscopy images sourced from publicly accessible platforms such as RxRx3 and JUMP-CP. This innovative training method facilitates revolutionary advancements in extracting valuable insights from complex microscopy data.
The Importance of Large-Scale Cell Microscopy
Modern research increasingly relies on large-scale cell microscopy assays to unveil essential biological processes and discover novel drug candidates. Historically, laboratories have utilized traditional image segmentation and feature extraction software that requires significant manual tuning for different assays. Such methods often demonstrate reduced effectiveness in adapting to changes in experimental conditions. In contrast, OpenPhenom-S/16 is anticipated to revolutionize the field by enabling researchers to switch to this pre-trained model, enhancing their workflows without needing further adjustments or training.
Improvements Seen with OpenPhenom-S/16
Preliminary results indicate that OpenPhenom-S/16 significantly improves the recall of known biological relationships when assessed against the StringDB database on the public JUMP-CP cpg0016 dataset. This newfound capability offers researchers a streamlined path to derive insights from biological data, ultimately facilitating rapid advancements in drug discovery research.
Collaboration with Google Cloud
Shweta Maniar, the global director of Life Science Strategy & Solutions at Google Cloud, expressed enthusiasm about this collaboration. She emphasized the remarkable scalability of Google Cloud's Vertex AI platform, highlighting how OpenPhenom-S/16 can automate the classification of cellular structures, detect patterns and anomalies, and extract important information from complex datasets. By democratizing access to this advanced technology, Recursion and Google Cloud aim to empower researchers globally to make critical discoveries in biotechnology.
Recursion's Commitment and Future Goals
As a leader in the biopharmaceutical sector, Recursion has developed models for both internal and commercial use, enhancing its proprietary pipeline. The company has previously showcased the effectiveness of its proprietary Phenom models, which have been pivotal in its ongoing research initiatives.
Introducing the RxRx3-core Dataset
In conjunction with the launch of OpenPhenom-S/16, Recursion also released the RxRx3-core dataset, designed specifically for the research community. This dataset encompasses labeled images representing 735 genetic knockouts and 1,674 small-molecule perturbations. The RxRx3-core dataset aims to bridge the gap in mapping drug mechanisms, an essential element in utilizing high-dimensional biological data effectively.
Accessing OpenPhenom-S/16
Researchers and life science professionals interested in utilizing OpenPhenom-S/16 can access it through Google Cloud. Furthermore, they can find detailed information regarding the RxRx3-core dataset, which plays a fundamental role in the scientific exploration of drug discovery.
About Recursion
Recursion (NASDAQ: RXRX) stands at the forefront of the TechBio industry, with a focus on decoding biological complexities to industrialize the drug discovery process. Central to this mission is the Recursion OS, an innovative platform that integrates advanced technologies, fostering one of the largest proprietary biological and chemical datasets worldwide. By utilizing sophisticated machine-learning algorithms, Recursion extracts an enormous array of searchable relationships, propelling significant advancements across the fields of biology and chemistry.
Contacting Recursion
For those wishing to connect with Recursion or seeking further information on their groundbreaking work, the company can be reached via the following:
Media Contact: Media@Recursion.com
Investor Contact: Investor@Recursion.com
Frequently Asked Questions
What is OpenPhenom-S/16?
OpenPhenom-S/16 is a foundation model developed by Recursion, publicly available for enhancing microscopy analysis and accelerating drug discovery.
How does OpenPhenom-S/16 improve drug discovery processes?
This model enables researchers to analyze complex microscopy data more effectively without requiring manual tuning, thus speeding up the drug discovery timeline.
What datasets are utilized to train OpenPhenom-S/16?
OpenPhenom-S/16 has been trained on over three million microscopy images, utilizing publicly accessible datasets like RxRx3 and JUMP-CP.
Who can access OpenPhenom-S/16?
The model is publicly accessible to researchers and life science companies through Google Cloud’s Vertex AI Model Garden.
What is the RxRx3-core dataset?
The RxRx3-core dataset includes labeled images of genetic knockouts and small-molecule perturbations, designed to help researchers investigate drug mechanisms.
