This presentation surveys how Flagship Pioneering’s portfolio companies use cloud technologies to scale AI/ML models. We’ll examine architecture patterns, strategies, and lessons learned, showcasing methods for scaling compute and pipelines. Additionally, the talk will highlight emerging trends in cloud and AI/ML, offering insights into how these innovations are shaping the future of biotech research. Attendees will gain practical strategies for optimizing cloud environments to drive AI/ML growth in biotech.

This presentation will detail challenges and solutions in implementing large-scale human genetics to support sequence-based precision medicines like RNAi therapeutics. Key topics will include data management, cost effectiveness, and navigating ever more stringent data security standards. The talk will highlight innovative strategies and technologies that Alnylam put in place that paved the way for the development and delivery of novel RNAi therapeutics to patients in need. Human genetics is at the heart of drug discovery at Alnylam. Effective, large scale human genetics is a data management and computational challenge. The loss of trust and the growth of human genetics projects as data.

Flexible Architecture for Machine Learning for Genomics: This talk will explore the architecture for developing, deploying, and scaling machine learning models in genomics. It will emphasize the importance of creating a flexible infrastructure that can accommodate a range of ML workloads for drug development. Key topics will cover optimizing model training, managing datasets, and efficiently utilizing compute resources to meet different workload demands while ensuring cost-effectiveness and scalability.

Drug discovery is a collaborative effort, with AI/ML emerging as a transformative partner. Beyond its role in everyday life, AI accelerates drug discovery by streamlining processes and providing innovative solutions. This talk explores how ML is integrated into Roche's in-house developed product suite, enabling teams to efficiently manage small molecules through profiling cascades with automation features. By harnessing ML, the platform facilitates the skipping of assay requests, optimizing the workflow. Described as "Lab-in-the-Loop," this mechanism enriches Roche's digital ecosystem in Discovery and Early Development. Join us to discover how an iterative ML model is revolutionizing drug discovery.

In this talk, we will show how image analysis can be taken from research level software to high quality and production ready systems. Attendees will learn how Jackson Laboratory transitioned research-level software into production-ready systems for scalable and durable image analysis. This presentation will feature biological image data, offer insights into new paradigms for cloud-based analysis, and appeal to those interested in scaling analysis of any type using distributed programming and cloud technologies.

The collaborative efforts within Takeda, supported by external vendors, led to the development and prototyping of One Research Digital—an ecosystem comprised of FAIR data lakes. During the design phase, we adhered to the FAIR principles and utilized a data lake infrastructure with a flexible data storage schema as the system's cornerstone. These advancements enable us to enhance AI-augmented analysis, modeling, and reporting across preclinical and translational sciences.

Reaction-based virtual libraries have expanded our access to chemical spaces of billions of virtual molecules. Searching such spaces for molecules of interest can be performed in many ways, but they can usually be parallelized at multiple levels. We describe a virtual screening infrastructure that uses Nextflow to parallelize a Genetic Algorithm Virtual Screening (GAVS) coupled to computational chemistry primitives, such as shape matching and docking, using AWS Lambda. Finally, we present some benchmark results highlighting the efficiency and accuracy of the method, along with the performance gains achieved by virtue of cloud-based parallelization.

As AI and machine learning transform biopharma R&D, building a scalable, cost-efficient, and compliant cloud infrastructure is essential for accelerating innovation. We have embraced a fully cloud-native approach to power AI/ML workloads, enabling seamless access to data, high-performance compute, and secure collaboration. This session will explore key architectural decisions, trade-offs considered, and best practices for optimizing cloud environments to support AI/ML at scale. Attendees will gain insights into leveraging cloud technologies to drive scientific breakthroughs while maintaining flexibility, security, and cost efficiency.

Since 2010, “Trends from the Trenches” has been a cornerstone of the Bio-IT program, delivering candid and occasionally blunt assessments of the most impactful and overhyped IT technologies in life sciences. This talk will provide a deep dive into computing, storage, cloud, data science, machine learning, and more, with a focus on supporting data-intensive science. Looking ahead, this talk will share forward-thinking predictions about emerging technologies and trends poised to shape the future of life sciences innovation, offering actionable insights for navigating the next wave of IT evolution.

Launching in 2026, a new AI supercomputer powered by Nvidia’s latest Rubin-generation GPUs will transform research at Oregon State University’s Huang Collaborative Innovation Complex. This mini talk highlights its capabilities, from accelerating protein structure prediction to advancing quantum simulations to something completely new and different. Learn how you can get access to this cutting-edge resource and drive innovation in life sciences and quantum computing simulations and discover opportunities to collaborate.

As the world's largest publicly available repository of raw sequence data, the Sequence Read Archive (SRA) plays a pivotal role in advancing public health and life sciences research. This presentation highlights state-of-the-art tools and strategies for managing and analyzing the SRA’s massive datasets, showcasing its impact on infectious disease surveillance, genomic epidemiology, and precision medicine. Discover how innovative informatics solutions are transforming raw data into actionable insights for global health challenges.

Many biologists who have spent years coding and thinking in terms of bioinformatics - protein and DNA sequence, genomics - are now engaging with machine learning, NLP, and LLMs. In this talk a bioinformaticist will talk about the many lessons learned and twists and turns encountered in these new fields. Topics will include new ways of thinking about computing with CPUs and GPUs, re-representing data, training, iteration and validation, version control and environments, new definitions of “pipeline”,  and coming face-to-face with prediction and statistics.