NEW YORK – Google is the latest of the major cloud platform service providers to make a formal foray into the world of genome analysis and artificial intelligence-driven genome interpretation.
The firm last week introduced Google Cloud Multiomics Suite and a companion product for drug discovery called Target and Lead Identification Suite. Launched in conjunction with the Bio-IT World Conference & Expo in Boston, the technology is meant to accelerate multiomic and multimodal data analysis and interpretation as well as to improve the success rate of drug development.
"Google was founded on the idea of how [to] bring more information to people to improve their lives at a vast scale," said Shweta Maniar, global director of life sciences strategy and solutions for Google Cloud. "We've extended that philosophy where we actually hope to improve people's lives by enabling pharma companies, biopharma, biotech, and public-sector organizations to … deliver on the promise of precision medicine."
Maniar talked of a "four-headed beast" of data that must be tamed to achieve precision medicine. "It requires an enormous amount of resources from an acquisition perspective, storage, distribution, and analyzing genomic data," she said.
"We need as much data as possible from as diverse of a pool as possible," Maniar added. She said that the Google Cloud Multiomics Suite is intended to help research organizations manage and perform large-scale analysis and stratification of processes including clinical trials, adding that workflows should be reproducible.
The Multiomics Suite ingests data and serves as a hub for data sharing and management. Working with Google Cloud partners, the software lets users bring in various open-source and proprietary analysis and interpretation tools to help make sense of the information.
The Google announcement closely followed news from Amazon Web Services, which also took advantage of Bio-IT World to introduce new features for its nascent Amazon Omics service, including a collection of "pre-built" workflows, support for graphical processing units (GPUs), direct data uploads through an application programming interface (API), and streamlined variant querying and analysis.
Maniar did not address Amazon Omics directly, but called traceability a "differentiator" for the Multiomics Suite, backed by Google Cloud's Vertex AI platform. She noted that Google is "really, really leaning in hard" on its AI capabilities in both new products.
Amazon Omics actually debuted in late November, giving AWS a six-month head start on Google Cloud.
Initial Google Cloud Multiomics Suite and Target and Lead Identification Suite partners include Form Bio, Nvidia, Epam Systems, Omnigen Medical Products, and data sciences firms Quantiphi and Max Kelsen.
Dallas-based Form Bio, a 2022 spinoff from Colossal Biosciences, is providing access to its computational life sciences platform via the new Multiomics Suite.
Colossal, the gene editing startup that has grabbed headlines with its ambitious plans to "de-extinct" long-disappeared species including the woolly mammoth, is a development partner and beta tester of Google's Multiomics Suite. That firm said in a statement that the software reduced whole-genome sequencing analysis time by 88 percent and cost by 52 percent compared to open-source tools.
Christopher Haas, Form Bio's head of partner engineering, said that his firm is applying large language models to multiomics data to support research in genetics, synthetic biology, and cell and gene therapies.
"We're trying to make it really easy for bench scientists to manage their data, run the workflows that are relevant to their research needs, and then analyze and visualize the output," said Haas, a former Google executive. "We could deploy a customer tomorrow into the platform."
Form Bio customers span the private and public sectors, but Haas noted that state and local public health agencies have turned to the company's technology for tasks including variant analysis for infectious diseases. He said that they might use a platform like Google's Multiomics Suite to figure out if a batch of COVID-19 test samples contains a new variant.
Form built its technology on Google Cloud, but it has a federated architecture that would be able to support customers who want to run the software on other clouds, including AWS or Microsoft Azure. The company does not currently have a relationship with Amazon Omics.
Haas said that the Multiomics Suite is "very broad," leaving room for Form's partnership with Google to expand. Form will be able to bring its FormSightAI module for biomanufacturing onto the new Google software platform to make in silico predictions of the manufacturability of a compound before a customer spends millions of dollars on actual manufacturing, according to Haas.
As part of the partnership, Form Bio and Google will be able to cross-sell multiomics services, a major boon for a startup.
Nvidia, another launch partner for the Google Multiomics Suite, also has partnerships with Form Bio and Colossal Biosciences, in addition to supplying graphics processing units to Google Cloud. For the Multiomics Suite, that firm is offering its Parabricks genomic analysis technology, according to Jason Fenwick, an Nvidia business development specialist in genomics.
Nvidia last week released version 4.1 of Parabricks, which features a tool for retraining Google's DeepVariant to suit individual datasets, allowing for more accurate variant calling. Fenwick said that will be a key benefit of integrating Parabricks with the Google Multiomics Suite.
DeepVariant is one piece of the AI that Google is integrating into its two new life sciences suites.
With the Multiomics Suite and the Target and Lead Identification Suite, Google is not handling any patient-specific data, known as protected health information (PHI) for the purposes of HIPAA in the US. "If there is going to be an occasion [to include PHI], the customer would control that access," Maniar said.
For this reason, Google Cloud is focusing on research rather than clinical practice with these two software suites. Maniar said that Google Cloud is open to adding more clinically focused features later on if that is the direction the healthcare industry is headed.
"These solutions can be part of the path to enable organizations [to improve] diagnosis and treatment of diseases, the prevention of diseases, and personalized treatment plans in the future," Maniar added.
Lukas Karlsson is a cloud consultant in the Boston area with a long history of development on Google platforms, including while at the Broad Institute.
Karlsson, who has spoken at Google-sponsored events, said that the initial partnerships represent a starting point, and some may or may not pan out. What is clear to him is that Google needed to respond to the market.
"Everyone is under the utter necessity of stitching some kind of AI functionality into every product," Karlsson said. "That's clearly the background of the timing."
Karlsson said that this announcement might be an easy way for Google to add AI to life sciences very quickly and give scientific users a taste of what the technology is capable of. He also suggested that Google might be "kind of afraid to let it loose on people because people aren't ready for what it's able to do yet in general."
Karlsson said that Google's AI is much more powerful than what the firm is presenting at the launch of these two software suites. "I think it is the case with the genomics features … that they are very much just dipping a toe in the water," he said.
Karlsson referenced Blake Lemoine, a software engineer whom Google fired a year ago after he internally and later publicly raised concerns that the firm had unreleased AI technology that was sentient.
Karlsson's hope is that Google is starting slowly because it plans on adding increasingly advanced features in the future.