Model matches the accuracy of leading 1.2 billion-parameter DNA language models while using only 172 million parameters

Inocras, a bioinformatics-led company harnessing the power of whole genome data and proprietary analytics to deliver curated insights that advance precision health, announced that “DNAChunker: Learnable Tokenization for DNA Language Models,” a joint research paper with the Korea Advanced Institute of Science and Technology (KAIST), has been accepted for paper publication at the International Conference on Machine Learning (ICML) 2026.

The paper introduces DNAChunker, a learnable adaptive tokenization approach for DNA language models that dynamically segments genomic sequences into biologically meaningful, variable-length units. Unlike conventional DNA language models that process genomic sequences using fixed-size or externally defined segments, DNAChunker learns how to group genetic code based on biological context, enabling more accurate and efficient representation of complex genomic patterns.

DNAChunker achieves state-of-the-art performance while matching the accuracy of leading 1.2 billion-parameter DNA language models with only 172 million parameters, making it more than seven times smaller. By reducing model size while preserving performance, DNAChunker may help make advanced genomic AI models more practical for large-scale research, translational discovery and future clinical applications.

“DNA language models depend heavily on how genomic sequences are represented before they are interpreted by AI,” said Wonchul Lee, CIO at Inocras and co-lead of the paper. “By replacing rigid tokenization with a learnable approach, DNAChunker provides a more precise and efficient foundation for downstream genomic modeling.”

Also Read: Tempus Introduces ‘Preview’: Bridging the Critical Time Gap Between Diagnostic Order and Definitive Results

“Our ICML acceptance marks a major milestone for Inocras’ Cancer Foundation Model, developed in collaboration with KAIST and trained on thousands of whole genomes from diverse cancer types,” said Jehee Suh, CEO of Inocras. “DNAChunker provides the biologically informed genome representation layer underlying that broader vision, helping foundation models move beyond pattern recognition toward clinically meaningful cancer interpretation. Together with KAIST, we are advancing the core technologies needed to make whole-genome AI more accurate, efficient, and scalable.”

KAIST led foundational algorithm design, model implementation and validation, while Inocras contributed large-scale computational resources, key technical ideas and validation efforts to align the model with practical and clinical applications.

“DNAChunker shows that sequence representation is a central challenge in building effective DNA language models,” said Prof Sungsoo Ahn and Insu Han from KAIST and corresponding authors of the paper. “Our collaboration with Inocras helped connect advanced AI methodology with the scale and practical requirements of whole-genome analysis.”

Source: Businesswire

The post Inocras and KAIST Launch DNAChunker for DNA Language Models appeared first on ITDigest.

Also Read: Catalent Launches Qai to Transform Manufacturing QA

In fact, by switching to AI-powered simulations in virtual mode instead of experiencing structural biology with the traditional, labor-intensive experimental methods, this collaboration structure shall Really reduce the preclinical period, allow more efficient validation of the targets and finally pave the way for the design of completely new, highly potent therapeutics. Thanks to a state-of-the-art secure data infrastructure, this implementation brings for Takeda’s use of leading-edge machine learning setups while always respecting the strictly confidential and intellectual rights of the target data and molecules. The cooperation achieving such a high degree of technological advancement and industrialization of ultra-modern AI models in the pharmaceutical deep tech research field is a big step forward and So going beyond merely isolated proofs-of-concept to a fully standardized and scalable engine for autonomy in molecular discovery.

Read More: Boltz Announces Collaboration with Takeda to Deploy Frontier Biomolecular AI Models Across Takeda’s Research Organization

The post Boltz Partners with Takeda to Deploy Frontier Biomolecular AI Models appeared first on ITDigest.

Also Read: SandboxAQ Integrates Quantitative AI Models With Claude to Accelerate Scientific Discovery 

“The launch of Qai represents an important milestone in how Catalent is applying advanced technologies, including AI, across our operations to improve consistency, accelerate insight and enable better decision-making. Innovative AI solutions like Qai strengthen the quality of our operations and better support the teams delivering critical therapies to patients around the world.” Emphasizing the industrial value of high-trust data governance, Todd Mersch, General Manager, U.S. Life Sciences and MedTech, Microsoft, added: “Qai reflects what we see as an important application of AI in life sciences: enhancing already robust data analytics and governance to deliver meaningful patient impact. By strengthening oversight and consistency across manufacturing processes, Qai supports Catalent’s commitment to delivering for customers and the patients they serve, helping to transform lives.” Ultimately, this roll-out marks a significant step in Catalent’s broader, ethically grounded digital transformation strategy, empowering global operations teams to minimize risk and advance a “Patient First” culture.

Read More: Catalent Launches Qai to Reimagine Quality Assurance for its Manufacturing Services

The post Catalent Launches Qai™ to Transform Manufacturing QA appeared first on ITDigest.

By integrating these new gene expression tools, laboratory automation partnerships, and specialized analysis software, QIAGEN aims to accelerate the adoption of digital PCR across large-scale molecular biology research and regulatory compliance environments.

“Gene expression represents one of the largest application areas in molecular biology and a significant opportunity for digital PCR,” said Thierry Bernard, CEO of QIAGEN. “By expanding the QIAcuity ecosystem with new assays, enhanced multiplexing capabilities and workflow solutions, we are helping customers apply digital PCR to a broader range of research and biopharma applications.”

Also Read: Tempus Introduces ‘Preview’: Bridging the Critical Time Gap Between Diagnostic Order and Definitive Results

Expanding the QIAcuity Digital PCR Framework

The updated QIAcuity portfolio delivers key operational updates across four core segments of the biotechnology and life sciences market:

1. Advanced Gene Expression Solutions for dPCR

QIAGEN intends to expand its testing commercial lines later in 2026 by rolling out new QIAcuity Gene Expression Assays. The upcoming products are optimized to support precise expression profiling across human, mouse, and rat research models. Additionally, the developer plans to launch the QIAcuity OneStep High Multiplex Probe PCR Kit, a solution designed to maximize sample utility by enabling complex, high-order multiplex targets to run simultaneously inside a single nanoplate partition.

2. Enhanced Quality Control for Cell and Gene Therapy (CGT)

To support the stringent manufacturing regulations governing biopharmaceutical production, QIAGEN is building out its residual DNA testing suite. The expansion includes new quantification kits optimized for an array of producer cell systems. These tools enable biopharma operations to achieve reproducible, high-precision analytics during process development, batch release testing, and host-cell impurity monitoring.

3. Standardized Laboratory Automation Integration

Recognizing the throughput demands of industrial and clinical research laboratories, QIAGEN has partnered with Hamilton to automate the initial handling of QIAcuity nanoplate configurations. By transitioning fluid transfer, loading, and setup workflows to automated liquid handling systems, the collaboration significantly increases daily laboratory productivity, optimizes reagent usage, and minimizes the human errors inherent to manual pipetting.

4. Automated Analysis and Compliance Reporting via QIAcuity Software 3.5

Concurrently, the upcoming release of QIAcuity Software 3.5 introduces an updated data analysis layer. The software features pre-configured analysis templates and automated reporting frameworks built to standardize data interpretation across distributed laboratory setups. The platform upgrade streamlines verification speeds, making it easier for biopharma networks to preserve data integrity and maintain strict regulatory compliance across high-volume experimental runs.

Driving Data Efficiency via Nanoplate Partitioning

The QIAcuity platform distinguishes itself from legacy infrastructure by utilizing microfluidic nanoplates to disperse a biological sample into thousands of individual, isolated partitions. The system then executes and reads the target reactions simultaneously, allowing it to quantify even faint, low-abundance genetic signals with absolute accuracy. By combining sample partitioning, thermocycling, and optical imaging into a single unified instrument, the platform condenses traditional digital PCR workflows from six hours down to just two.

The comprehensive portfolio updates and software enhancements are scheduled to roll out across global biopharma and life sciences markets throughout the remainder of 2026. Laboratories can explore platform specifications, review automation integration protocols, and access relevant application notes via the official QIAGEN digital portal.

Source: BusinessWire

The post QIAGEN Broadens QIAcuity Digital PCR Portfolio to Drive Biopharma Innovation appeared first on ITDigest.

The initial days following an advanced cancer diagnosis are critical for strategic treatment planning, yet clinicians have traditionally been forced to operate in an information vacuum while awaiting comprehensive genomic profiling results. Tempus Preview fundamentally redefines this diagnostic timeline. By combining Tempus’ multimodal data and advanced AI capabilities applied directly to the earliest touchpoints of the laboratory workflow, Tempus equips care teams to access early, clinically significant, information that can help inform complex decisions for patients and shorten the time between receipt of final molecular results and implementation of a personalized treatment plan.

At launch, Tempus Preview will focus exclusively on high-impact biomarkers where early insights can be critical, including:

• Surfacing patients more likely to harbor microsatellite instability (MSI-H), a biomarker linked to improved response to immune checkpoint inhibitors and potential hereditary risk factors, in colorectal, endometrial, prostate, and esophagogastric cancers.
• Predicting EGFR mutations in non-small cell lung cancer (NSCLC), a biomarker that infers response to targeted therapy, but often lacks response to frontline immunotherapy.
• Highlighting increased probability of potential rare, yet clinically significant FGFR fusions in hepatobiliary and bladder cancers, that may indicate potential response to targeted therapy and improved patient prognosis if gene fusions are present.

Also Read: Benchling and Baseten Partner to Bring AI Inference to Biotech R&D

Shortly thereafter, Tempus Preview will expand to other critical biomarkers.

Tempus Preview’s biomarker predictions are powered by Paige Predict, an advanced AI model that analyzes standard H&E images to provide genomic insights. Paige Predict, trained on millions of slides, has been validated for clinical use as part of Tempus’ laboratory-developed test.

“At Tempus, our unique combination of a diagnostic lab and an advanced data platform enables us to build AI models powered by our unparalleled depth of real-world data,” said Eric Lefkofsky, Founder and CEO of Tempus. “That foundation creates a powerful flywheel: every insight strengthens our models, and every model helps generate more clinically meaningful insights for providers and patients. Tempus Preview brings this intelligence directly into the clinical workflow, delivering early, clinically relevant information within one day of sample receipt, which for many patients can mean the difference in how they are treated. This is our AI flywheel in action: transforming complex information into timely insights, delivered to physicians when they need them most.”

Source: BusinessWire

The post Tempus Introduces ‘Preview’: Bridging the Critical Time Gap Between Diagnostic Order and Definitive Results appeared first on ITDigest.

Built on Biocytogen’s proprietary RenMice^® platforms, RenSuper establishes a closed-loop antibody discovery engine combining in vivo immune repertoires, AI-driven candidate selection, automated experimental validation, and scalable manufacturing infrastructure. The platform supports the discovery and development of monoclonal antibodies, bispecifics, multispecifics, antibody-drug conjugates (ADCs), VHHs, and other advanced therapeutic modalities, significantly reducing the time and risk associated with antibody discovery.

RenSuper Workstation: AI-Powered One-Click Target-to-Lead Therapeutic Antibody Discovery

RenSuper Workstation introduces a new paradigm for one-click therapeutic antibody discovery, transforming antibody discovery from a project-based workflow into a searchable and programmable system. The platform enables rapid identification of high-quality therapeutic candidates with strong developability, functional relevance, and translational potential.

Proprietary Fully Human Antibody Library

The platform provides off-the-shelf access to fully human antibody sequences against more than 1,000 validated targets, complete with data packages to support more confident decision-making.

AI-Powered Candidate Selection

RenSuper’s AI models are trained on more than 100 million biologically validated antibody sequences, enabling efficient candidate identification against structurally and biologically challenging targets.

Using Biocytogen’s RenNano^® fully human HCAb platform as an example, the AI-powered screening workflow achieved an average positive hit rate of 46%, with peak hit rates reaching 98%. The workflow also enriched candidates with superior developability profiles and delivered an 82% success rate for high-purity (≥90%) Fc-format antibody expression.

Also Read: QIAGEN Launches QIA Agent to Bring AI-Driven Scientific Guidance into Research Workflows 

Experimentally Validated Antibody Sequences

All AI-selected antibody sequences undergo experimental validation, including binding, affinity, and specificity characterization, ensuring reproducibility, reliability, and downstream development readiness.

One-Click Search & Design

The platform enables one-click filtering and comparison of antibody candidates by target, epitope, and physicochemical properties. An integrated antibody design platform further supports rapid sequence assembly and flexible format design across multiple therapeutic modalities.

RenSuper High-Throughput Automation Center

Beyond intelligent discovery, RenSuper integrates industrial-scale automation to establish a scalable, closed-loop antibody development engine.

The newly launched automation center enables high-throughput experimental validation of AI-selected antibody candidates with minimal manual intervention, significantly accelerating the transition from molecular design to functional evaluation.

The fully automated platform spans the entire protein production workflow, including bacterial inoculation, plasmid extraction, transfection, fed-batch cultivation, and purification.

The system consistently delivers antibody yields averaging 50-100 mg, with throughput of up to 800 samples per day and over 5,000 samples per week, transforming antibody validation from a labor-intensive process into a scalable industrial workflow.

“We believe the future of antibody discovery will be driven by speed, scale, and data-driven intelligence,” said Dr. Yuelei Shen, Founder and CEO of Biocytogen. “With RenSuper, we are building an intelligence-native system that makes antibody discovery more intelligent, scalable, and efficient. Our goal is to reduce the time and complexity of early-stage discovery, accelerate the development of innovative therapeutics, and bring new medicines to patients faster.”

Together, RenSuper Workstation and the RenSuper High-Throughput Automation Center establish a next-generation target-to-lead ecosystem powered by biologically validated immune repertoires, artificial intelligence, and industrial-scale automation.

This milestone further reinforces Biocytogen’s vision of becoming the “Global Headstream of New Drugs” by transforming therapeutic antibody discovery into a more intelligent, scalable, and programmable process for biologics innovation worldwide.

Source: BusinessWire

The post Biocytogen Launches AI-Powered RenSuper™ Platform and Industry-First Fully Automated Antibody Discovery Infrastructure appeared first on ITDigest.

The QIAGEN team believes that QIA Agent allows for getting information on science, receiving product recommendations, accessing relevant data on workflow planning, accessing technical documentation, and obtaining other assistance in managing experiments using one conversational interface. With QIA Agent’s help, scientists will be able to plan their experimentations, get recommendations on choosing products for it, compare solutions available, obtain access to protocols, and get more information on any technical issues.

It should be noted that QIA Agent operates as part of the whole QIAGEN digital ecosystem and works in connection with such applications as Experiment Configurator, Product Availability Checker, and Order Status Checker. As QIAGEN’s team claims, the main purpose of this application is to make the way from the beginning to completion of an experiment easier for a researcher.

Nitin Sood, Senior Vice President and Head of Product Portfolio & Innovation at QIAGEN, noted that researchers are facing increasing scientific complexity, growing volumes of data, and expanding workflow choices. The new platform is designed to address these challenges through a unified AI-powered experience that streamlines access to scientific and operational information.

Also Read: Virtualitics and OpenAI Join Forces to Develop Next-Generation AI Readiness Solutions

AI Expands Its Role in Scientific Research and Laboratory Technology

The release of QIA Agent marks an emerging trend within the life sciences and tech sectors. Typically, researchers would be required to engage with a variety of software environments, databases, technical documentation, and searching procedures to navigate their workflow. However, the emergence of generative artificial intelligence and agentic systems now allows organizations to build intelligent assistants that will not only be able to understand the context of their work but synthesize relevant data and assist users with completing complex scientific experiments.

For the IT sector, the emergence of such solutions signifies another step towards the growing convergence of AI technologies with scientific computing, cloud research infrastructure, and corporate knowledge management. Indeed, the emergence of specialized software with built-in AI assistants can be expected to lead to a growing need for software solutions based on AI infrastructure and data platforms, knowledge graphs, and domain-specific AI capabilities.

From a broader perspective, the emergence of AI assistants is part of an ongoing process whereby artificial intelligence technologies are being incorporated into scientific research. It is quite obvious that vendors in the health, biotech, and pharma industries will seek to enhance their capabilities by developing AI solutions to facilitate scientific research.

Business Impact and Industry Outlook

For companies working in life sciences, diagnostics, pharmaceuticals, and other research-oriented industries, an artificial intelligence-based scientific assistant would make a major contribution to the optimization of the workflow and the increase of productivity. Often, researchers dedicate a considerable amount of time to the search for protocols, technical descriptions, products, and the coordination of work processes. Using such an intelligent system would allow companies to cut down administrative costs and let researchers focus on research.

Additionally, with such an assistant, a company might achieve faster research times through easier and more effective access to scientific knowledge. With better information access and workflow organization, researchers will be able to conduct experiments faster and more efficiently.

However, the increased use of AI in scientific laboratories raises important issues related to the explanation of actions performed, especially since they can have both scientific and economic consequences. Therefore, it is essential that the assistance provided by AI be based on human input and expertise.

The Future of AI-Connected Scientific Workflows

QIAGEN’s QIA Agent launch highlights a broader transformation underway across the scientific technology landscape. As AI agents become increasingly capable of understanding scientific context and supporting complex workflows, research platforms are evolving from static software tools into intelligent assistants that actively guide users through discovery and operational processes.

For the IT industry, this development signals a future where AI-powered domain-specific assistants become standard components of enterprise software ecosystems. Organizations that successfully integrate these technologies may gain advantages in research productivity, operational efficiency, and innovation speed as AI continues to reshape the future of scientific computing and digital transformation.

The post QIAGEN Launches QIA Agent to Bring AI-Driven Scientific Guidance into Research Workflows appeared first on ITDigest.

Between 2020 and 2025, the number of new scientific AI models released annually grew from 28 to more than 380. These models are now standard in R&D workflows, but the compute layer hasn’t kept up. Drug discovery is bursty by nature: teams wait on the physical lab, data comes in waves, then need to run 100,000 predictions in a few hours before going quiet for days. For most computational teams, that plays out as HPC queues with multi-week backlogs, GPU reservations sitting idle between data collection cycles, and predictions rationed during active campaigns.

Benchling Inference is built on the Baseten Inference Stack, a tightly integrated combination of a high-performance runtime (custom kernels, speculative decoding, KV cache optimizations) and inference-optimized infrastructure spanning 15+ cloud providers, with cold starts in 5–10 seconds. Benchling adds a biotech layer on top with pre-configured defaults for scientific models and deployment options for organizations with strict data residency requirements. By aggregating demand across the industry, Benchling also brings better economics to biotech startups.

Also Read: Owkin to Build AI Agents as Part of a Multi-Year K Pro License Agreement With AstraZeneca

With Benchling Inference, scientists can deploy third-party models or serve internal models built on their own experimental data from a unified compute environment. For teams with data sovereignty requirements, the Baseten Inference Stack runs identically in Baseten Cloud, inside a customer’s virtual private cloud (VPC), or a hybrid of both so predictions never have to leave their environment. Computational scientists working in Jupyter notebooks or via SDK can call inference directly through Benchling.

“Biotech has entered a new era where AI models trained on proprietary experimental data could unlock breakthroughs that weren’t possible before. The bottleneck has been infrastructure and biotech research labs should not have to become GPU experts to run frontier models on their data. By partnering with Benchling, we bring six years of inference expertise directly into the environments where the science happens” said Amir Highighat, CTO & Co-Founder of Baseten.

“Access to compute is becoming a strategic advantage. But we hear from computational scientists that getting inference to work in drug discovery is harder than it should be; workloads are bursty, the data is sensitive, compute costs are too high,” said Ashu Singhal, co-founder and President of Benchling. “We’ve been running Baseten internally for Benchling’s Model Hub and learned a lot about tailoring inference for drug discovery. Now we want customers to have the same access.”

Source: PRNewswire

The post Benchling and Baseten Partner to Bring AI Inference to Biotech R&D appeared first on ITDigest.

Also Read: IBM Marks a Decade of Quantum Computing on the Cloud, Driving Global Innovation

This model helps speed up catalyst discovery by quickly calculating adsorption energy. This is crucial for evaluating catalyst performance in sustainable fuels, hydrogen, and recycling plastics. SandboxAQ also revealed plans to bring additional pharmaceutical-focused models, including AQPotency and AQCell, to Claude in the near future. “Now, researchers can access frontier physics-based models directly inside the AI tools they already use, with no additional infrastructure, code or barriers,” said Jack D. Hidary, CEO of SandboxAQ. Industry experts noted that combining conversational AI with quantitative scientific models could dramatically shorten discovery timelines and improve accessibility for researchers across scientific disciplines.

Read More: SandboxAQ Integrates its Quantitative AI Models with Anthropic’s Claude via MCP

The post SandboxAQ Integrates Quantitative AI Models With Claude to Accelerate Scientific Discovery appeared first on ITDigest.

Under the three-year licensing agreement, Owkin will lead the end-to-end development of AI agents to run on K Pro, integrated within AstraZeneca’s IT infrastructure and decision workflows. The new agents’ functionality is intended to help AstraZeneca’s decision-making teams access timely, data-rich insights for complex competitive intelligence questions, reducing reliance on manual analysis within established governance, security, and enterprise standards.

Also Read: Sapio Sciences Brings Claude Cowork to the Lab

“At Owkin, we believe the future of the pharmaceutical industry is agentic,” said Thomas Clozel, CEO and co-founder of Owkin. “Our experience, multimodal data, and agentic infrastructure allows us to build various complex agents supporting our pharmaceutical partners, including competitive intelligence agents to support quick decisions by executives.”

This new agreement builds on Owkin’s previous work with AstraZeneca to develop an AI gBRCA pre-screen solution for breast cancer. Ongoing results from this project were published at ESMO where the BRCAura RUO solution was found to rule out approximately 40% of patients unlikely to carry gBRCA mutations with a high sensitivity of 93%¹. This work now continues at Waiv, the recent spin-out of Owkin’s diagnostic division, while Owkin continues to focus on its core aim – developing biological artificial superintelligence to understand complex biology.

Source: BusinessWire

The post Owkin to Build AI Agents as Part of a Multi-Year K Pro License Agreement With AstraZeneca appeared first on ITDigest.