The Bio-IT World Conference & Expo 2024 (April 15-17) continued in Boston on Tuesday, April 16, and featured presentations in ten Bio-IT focused conference tracks: AI for Drug Discovery and Development; AI for Oncology, Precision Medicine, and Health; Bioinformatics; Cloud Computing; Data Management; Data Science and Analytics Technologies; Generative AI; Modern Data Platforms and Storage Infrastructure; Pharmaceutical R&D Informatics; and Software Applications and Services. The theme of this 23rd annual conference is “Building a Global Network for Precision Medicine.” The event has attracted nearly 3,000 attendees (in-person and virtual) from 30-plus countries around the world. A brief description of the ten conference tracks and a discussion of certain of Tuesday’s sessions are below, together with a discussion of the morning plenary session “Unveiling Tomorrow’s Possibilities: Embrace the Power of Digital Twins in Cancer Care and Research.”
CONFERENCE TRACKS
The ten Bio-IT focused conference tracks are briefly described here.
AI for Drug Discovery and Development
Leverage Artificial Intelligence and Machine Learning for Optimal Speed and Efficiency in Advancing Drug Discovery and Development Pipelines
The AI for Drug Discovery and Development track delves into the opportunities and obstacles faced by biopharma entities as they leverage the capabilities of artificial intelligence and machine learning technologies to optimize and expedite the entire drug discovery and development process, spanning from the initial stages to practical implementation. Speakers will examine AI’s role in reshaping our comprehension of diseases and target identification, approaches that blend AI with human expertise to identify and validate targets, and how it can elevate the field of chemical drug design and precision medicine. Additionally, we’ll explore how AI and machine learning initiatives measure up against the established and proven methods of drug discovery leading to successful market launches.
AI for Oncology, Precision Medicine, and Health
Enhance Accessibility and Exploration of Comprehensive Multiomics Real-World Data
Accessing robust multimodal real-world data (RWD) sets in oncology can often be challenging, as they are frequently compiled with incomplete genomic and clinical information. This lack of comprehensiveness hinders the integration of tumor biology with clinical practices. The AI for Oncology, Precision Medicine, and Health track explores technology tools and platform solutions to facilitate the application of artificial intelligence in disease pathology. What are the nuanced complexities of the problem at hand before deploying these technology tools? What data resources are accessible to ensure effective utilization? Throughout the track, speakers present best practice use cases that address these critical questions, ultimately aiding in harnessing information for the seamless transition of translational research findings into diagnostic support and clinical applications. A diagnostic provider that adopts a comprehensive patient profiling approach along with user-friendly tools for data linkage and analysis would provide immense value to the research community.
Bioinformatics
Harness Computational Tools and Techniques to Transform Extensive Data into Intelligent Insights, to Advance Pharma R&D
The Bioinformatics track convenes leading experts who will share case studies highlighting the use of computational resources and tools to conduct comprehensive analyses of multimodal data spanning genomics, transcriptomics, proteomics, metabolomics, and other data types, such as imaging. This data integration is aligned with clinical applications, aiming to transform big data into actionable insights for real-time support in disease prevention, prognosis, diagnostics, and therapeutics. As the volume of data generated for disease treatment and prevention continues to surge, bioinformatics technologies, tools, and methodologies remain pivotal in converting data into valuable knowledge, addressing unspoken medical needs. Additionally, the track explores topics like innovative single-cell data analysis methods, the synergy between bioinformatics and data science/AI in advancing research, and the application of deep learning in bioinformatics for pharmaceutical R&D.
Cloud Computing
Enable Collaboration and Drive Better, Faster Analytics Using Cloud Infrastructure and Applications
Adoption and deployment of cloud technologies is no longer an aspiration but a necessary mandate to enable digital transformation. There is simply no other way to store, manage, analyze, and share the massive amounts of data being collected to drive advancements in precision medicine. However, there are countless choices that must be made to determine the best path for your organization. Through case studies and best practices, the Cloud Computing track explores options and provides guidance on how best to determine the right cloud or hybrid infrastructure and applications to advance R&D, enable collaboration and innovation, and support flexibility to stay abreast of technological advances driving precision medicine.
Data Management
Leverage Data for Value Creation in Life Sciences
With the growing demand for computing power among life science researchers and those addressing big data challenges, it is essential for data storage infrastructure to possess scalability to efficiently manage billions of data points and files. The primary challenge lies in the effective administration of data, ensuring its integration, accessibility, sharing, linkage, analysis, and maintenance to drive transformative change within the organization. Are data mesh and data fabric the true solutions to these challenges? How can one extract meaningful insights from data and its contents to generate value? The Data Management track delves into these inquiries and explores various topics, including FAIR data principles, data reuse, governance, literacy, data federation, and standards, as well as data curation and harmonization.
Data Science and Analytics Technologies
Tools and Methods for Extracting Insights and Value from Data to Advance Biomedical Research
The Data Science and Analytics Technologies track explores the tools, technologies, and programming languages employed by data scientists to extract additional insights and value from data. During presentations, we will discuss the significance of scalable platforms versus personalized data science assistance, the transformation into a data-driven organization, inventive strategies for data management and analytics, the art of identifying the genuine questions that require answers, the practical impact of data science, and the practical application of data science tools.
Generative AI
Harness Data Potential to Drive Innovation and Advance Biomedical Research
Each year, the exponential growth of data presents a tremendous opportunity for biopharma organizations to realize tangible benefits. However, despite the massive surge in data volume and diversity, traditional methods of data analysis and sharing have failed to keep pace, hindering organizations from fully harnessing this data’s potential. Organizations encounter formidable challenges when attempting to unify data from disparate sources due to siloed data repositories, varying data formats, and incompatible legacy systems, impeding the derivation of comprehensive insights. Generative AI holds immense potential in revolutionizing various facets of life sciences research and development, from accelerating innovation to enhancing data-driven decision-making and streamlining labor-intensive tasks. The accuracy and reliability of generated data must be rigorously ensured, and ethical and regulatory considerations must be heeded when applying generative AI in life sciences. The Generative AI track discusses these critical issues along with barriers to adoption, data strategies and platforms, use cases, and predictions.
Modern Data Platforms and Storage Infrastructure
Design, Deploy, and Oversee Data Storage Solutions Optimized for Optimal Speed, Performance, and Cost Efficiency
Is the challenge of managing your data growing daily? Do you possess a flexible and robust data management framework to effectively store, process, analyze, transfer, and safeguard vast data volumes in alignment with your organizational guidelines? Are you familiar with balancing availability and interoperability? What strategies exist for scalable distributed and federated data analytics? How are you navigating the discussions surrounding speed, performance, and cost trade-offs? Which vendors are the most suitable for your needs? How do you assess the strengths and weaknesses of technological solutions? Which data storage methods and types prove to be efficient? Considerable strides have been taken by pioneering organizations in advancing large-scale data management, encompassing storage platforms, integration and migration strategies, and governance. The Modern Data Platforms and Storage Infrastructure track explores these inquiries and shares best practices derived from these endeavors.
Pharmaceutical R&D Informatics
Drive Precision Medicine Through the Digitalization of Pharma R&D
The urgency surrounding generating, organizing, and analyzing data in the pharmaceutical industry has not waned. In fact, the desire to develop new technologies and to speed development of infrastructure and special projects in our new normal continues to grow at record speed. The Pharmaceutical R&D Informatics track explores digital transformation efforts and strategies for optimizing operations via new technologies and services to create an efficient informatics ecosystem while meeting scientific, business, and regulatory demands. We address trends in data science and analytics, particularly for target assessment, pharmacology prediction, and generative chemistry, as well as key projects around data products and finding better, cheaper ways to FAIRify data for these products. Large language models, such as ChatGPT, are also discussed, as well as ways AI and machine learning are driving innovation in R&D.
Software Applications and Services
Design Software Application Architecture to Add Value to All Ecosystem Stakeholders
The Software Applications and Services track explores how organizations are shaping their data strategies and scientific decision-making by optimizing release management, automated testing, and harnessing software architecture tools and development trends. As the landscape of life sciences and healthcare evolves, so do the realms of Apps and Services. What tools are at our disposal for crafting Apps and Services? To what extent is “personalized” software development tailored to a specific company beneficial or necessary? Software forms the cornerstone of innovation, and a well-designed software application architecture can enhance the value within an ecosystem of stakeholders. The track will feature case studies highlighting software tools that facilitate data manipulation, data analytics approaches, data methods, standards, transparency, efficiency, security, and cost-effective solutions.
PLENARY ADDRESS
“Unveiling Tomorrow’s Possibilities: Embrace the Power of Digital Twins in Cancer Care and Research” was the plenary address delivered by Caroline Chung, MD, Vice President and Chief Data Officer and Director of Data Science Development and Implementation of the Institute for Data Science in Oncology, MD Anderson Cancer Center. In her description, shesaid that the talk wouldexplore the transformative potential of digital twins in revolutionizing cancer care and research and provide insights into how digital twins can help deepen biological understanding, accelerate drug discovery, and personalize therapeutic strategies to optimize treatment outcomes for every individual. Amidst the exciting opportunities are the challenges that must be tackled to harness the power of digital twins to advance precision oncology, empower researchers and clinicians with unprecedented insights, and improve patient outcomes.
SAMPLING OF TUESDAY’S PRESENTATIONS
The wide variety of presentations in Tuesday’s ten Bio-IT focused tracks is indicated in the descriptions below.
“Machine Learning in Precision Medicine @AbbVie: Multiomics Perspective” was presented by Abhishek Pandey, Group Lead, Pharma Discovery, Abbvie. In his introduction, Pandey said that he would explore the integration of diverse data modalities, including genomics, clinical records, and imaging, using machine learning for a comprehensive understanding of complex diseases in preclinical and clinical contexts. While traditional multiomics analysis combines RNA, proteomics, and clinical data, this presentation highlights the innovative inclusion of imaging data, and enriching the multi-modal framework. Learn about practical applications of multiomics analysis, combining imaging and genomics data, and delve into exciting developments in radiogenomics and pathogenomics for a holistic understanding of disease mechanisms.
“Using Explainable AI for Patient and Endpoint Selection in Small Multiomic Clinical Trials: Examples from Phase II Studies in Multiple Indications” was presented by Mohammad Afshar, CEO, Ariana Pharma. He noted that identifying optimal patient selection criteria, endpoints, and dose remain important challenges in drug development, especially when starting from small cohorts of patients. Explainable artificial intelligence (xAI) using formal concept analysis (FCA) as implemented in the KEM platform enables effective analysis of heterogeneous datasets, including small cohorts. KEM systematically explores all variable associations, identifying criteria for patient stratification, surrogate efficacy endpoints, and optimal dose. Examples will be discussed during the presentation.
“Harnessing the Power of Generative AI: Unlocking Insights from Real-World Data for Data-Driven Decisions in Pharma R&D” was a presentation by Anu Sharma, Principal Scientist, Merck & Company, Inc. In her introduction, Sharma noted that real-world data (RWD) is revolutionizing the pharmaceutical industry by offering valuable insights beyond clinical trials. It allows for the assessment of drug effectiveness, patient outcomes, and safety in real-world scenarios. However, the heterogeneity of data types, varying degrees of data completeness, and incompatible legacy systems create significant challenges and opportunities to perform any scientific study. In this talk, we will discuss how Generative AI can address challenges to enhance data-driven decisions, help identify trends, correlations, and insights aiding researchers.
“Target to Lead: A Platform for Early Discovery Data Management” was presented by Rachana Ananthakrishnan, Executive Director, Globus, University of Chicago. She noted that a growing number of computationally intensive research activities require commensurate large-scale data management. In particular, high-resolution imaging instruments such as cryogenic electron microscopes require automation of data flows to increase throughput and researcher productivity, as well as to ensure the instrument remains highly utilized. Globus was initially established as cyberinfrastructure for managed file transfer and secure data sharing. We have grown Globus into a comprehensive platform for research data management that includes services for data description and discovery, protected data management, and automation. The Globus platform-as-a-service is increasingly used to easily build and execute automated data flows in this context. We will describe how the platform facilitates end-to-end automation of complex research flows, and will present scenarios from research universities and national
facilities that illustrate implementation of common use cases.
“Using AI and Machine Learning (AI/ML) to Power Predictive Drug Discovery” was presented by Ahmad Haider, Senior Director, Vertex Pharmaceuticals, Inc. In his introduction, Haider said that during the drug discovery process, it is standard practice to test a large library of compounds for their biological activity towards both the intended target molecule, as well as peripheral off-target molecules. This process can be very time- and money-consuming. Pharmaceutical organizations are now accelerating this process by employing predictive modeling approaches that are used to generate quantitative predictions of molecular activities in a fraction of the time. In this talk, we will discuss Vertex’s approach to building such a predictive modeling infrastructure using modern machine learning operations and management techniques. We have built a modern MLOps platform using AWS-native services such as Amazon SageMaker, AWS Batch, AWS Lambda, API Gateway, S3, and Fargate which are designed to deliver low latency predictions, separate the model management from model training concerns, and provide well-defined interfaces for various applications and users. This platform enabled us to train, deploy, and manage tens of thousands of activity prediction models which can be retrained daily or on-demand. We also provide a scalable, flexible model management experimentation infrastructure that provides a bedrock for collaboration and ultimately delivers optimized processes for testing small molecules.
“A First Look at National-Scale Multimodal Cancer Research in Practice” was the title of a presentation by Prabhu Arumugam, Director of Clinical Data and Imaging | Caldicott Guardian, Genomics England. In his introduction, Arumugam noted that Genomics England and its partners are currently building the world’s largest multimodal research platform for cancer. The platform uniquely brings together clinical WGS and healthcare data—from Genomics England’s work enabling the NHS to deliver precision medicine through the Genomic Medicine Service—with hundreds of thousands of matched pathology and radiology images being digitized by the National Pathology Imaging Consortium and made accessible through AWS. Join us to hear about the progress and challenges of making 100PB of data queriable in the cloud; a first look at the sort of insights this platform can provide; and how industry can use it and partner with Genomics England and its participants to advance new therapies.
“Long-Read Sequencing and Variant Prioritization of Unexplained Early-Onset Colon Cancer Pedigrees” was presented by Melissa de la Bastide, Division of Research, Cold Spring Harbor Laboratory. In her introduction, she said that the underlying cause for the increased prevalence of early-onset cancer in recent years remains elusive, and many cases do not show obvious drivers upon standard short-read testing. Long reads uncover thousands of variants that are missed by short reads. We describe the long-read germline sequencing of two early-onset colorectal cancer trios, and a family consisting of two brothers with testicular cancer, and their unaffected parents. We survey multiple aspects of genomic variation including SVs, SNVs, and methylation changes, and to prioritize variants for further study to aid in identification of events that may contribute to cancer predisposition.
“AbbVie Intelligence—Building a Suite of LLM-Enabled Tools for the Enterprise” was a talk by Brian Martin, Research Fellow & Head of AI, AbbVie Inc. In his introduction, Martin explained thatAbbiVie has developed the AbbVie Intelligence platform—an enterprise-scale suite of capabilities enabled by generative AI and large language models. It includes Chat (a GPT-powered chat assistant), Analyze (capable of summarizing or rewriting entire documents), Translate (machine translation, customized for AbbVie), and Ask a Source (which grounds LLM answers in trusted knowledge sources). This is the first step on a broader roadmap for enabling our teams to leverage generative AI safely and effectively. This talk shares valuable lessons learned on how to scale LLM services to the enterprise, as well as useful insights on the use of retrieval-augmented generation, vectorization, database query generation, and other ways to ground LLMs in real knowledge.
“Scalable Visual Analytics for Digital Pathology” was a presentation by Robert Krueger, Assistant Professor, New York University. In his introduction, Krueger said that with new imaging technology, cancer biology is undergoing a transformation into a digital era. AI has enabled the analysis of such images in unprecedented scale. However, human domain knowledge and intervention are still essential, as data acquisition and experimental analysis pipelines are rapidly evolving. He will present scalable visualizations and visual analytics interfaces that integrate machine learning to support biologists and pathologists in their workflows. The approaches help experts to filter and classify cells, find cellular interaction, and annotate cancerous regions. Future work will address the transition to multimodal, volumetric, and time-varying analysis.
“Data-Driven Drug Discovery: Machine Learning in Drug Discovery @AbbVie” was presented by Abhishek Pandey, Group Lead, Pharma Discovery, Abbvie. He noted that traditional drug discovery is known for its high costs, lengthy timelines, and frequent lack of success. Machine learning has the potential to positively impact any or all of these aspects, namely cost, speed, and success. He has the privilege of leading a global team at AbbVie that specializes in the application of machine learning to drug discovery. During this talk, he will delve into how AbbVie is addressing the challenges of drug discovery through the utilization of machine learning. He will share various use cases, both successful and unsuccessful, shedding light on what the future holds for the pharmaceutical industry. Listeners will gain insights into how to effectively scale drug discovery using machine learning and will discover actionable opportunities that can benefit us all. The presentation will provide a glimpse into the strategies and algorithms of machine learning currently in use in drug discovery, offering a forward-looking perspective on the direction in which our collective efforts are headed.
“Unbiased NGS Analysis: Introducing Pangenotyper” was the title of a presentation by Steven Van Vooren, Director, Product Marketing, Velsera, and Brendan Gallagher, Head of Business Development, Sentieon. They noted that traditional NGS analysis faces reference bias, hiding crucial genetic variations, especially in non-European populations. Pangenotyper utilizes pangenome references to eliminate bias and boost variant calling accuracy. Say goodbye to incomplete variant calling and welcome Pangenotyper as the seamless alternative, surpassing traditional methods in accuracy, efficiency, and cost-effectiveness for research and clinical applications.
“Generative AI: Hype, Hope, and Reality” was presented by Rishi Gupta, Associate Director, Data Science, Novartis. In his introduction, he said that in this talk, we delve into the hype, hope, and reality of Generative Artificial Intelligence (AI) in the realm of general drug discovery, Generative Chemistry, and retrosynthetic analysis. Our work explores how AI and machine learning (ML) models, particularly large language models (LLMs), can be effectively utilized for complex retrosynthesis processes and drug discovery tasks. We scrutinize the practicality of AI in predicting novel, viable chemical entities, and its role in accelerating the cycle times and reducing the cost of drug discovery. The presentation will offer a balanced perspective, analyzing both the potential benefits and limitations of AI in this context, including the challenges in validating and interpreting AI-generated output. The ultimate goal of our research is to demystify the role of AI in drug discovery and establish a clear understanding of its current capabilities and future prospects in the field.
“Large Language Models and Knowledge Graphs: Better Together” was a talk presented by Ted Slater, Managing Principal, Scientific Informatics Consulting,EPAM. He said that AI has evolved significantly since the first chatbot in 1964. Today’s large language models (LLMs) like OpenAI’s ChatGPT can generate accurate responses, but can also produce incorrect or “hallucinated” answers. Research shows LLM-generated queries perform better when combined with knowledge graphs (KGs) compared to relational databases. EPAM’s DIAL, an AI orchestration platform, leverages this to enhance decision-making in biomedical research, drug discovery, and patient care, driving innovation in these crucial fields.
“Introducing Luma – A New Multimodal Therapeutic R&D Paradigm” was presented by Christian Olsen, Associate Vice President, Industry Principal, Biologics, Dotmatics. In his introduction, Olsen noted that for years, the biotech sector has struggled with fragmented R&D solutions, hindering seamless integration across labs. This results in siloed teams, outdated tech adoption, and inefficiencies. Innovation demands integrated, multimodal approaches. Introducing Luma: a low-code SaaS platform revolutionizing R&D. Luma enables seamless integration with lab equipment and software, streamlining workflows. Learn how Luma accelerates R&D, shortening drug discovery timelines, and reducing costs.
“SaaS Application Architecture, Why Does it Matter?” was presented by Dan Schutzman, Senior SaaS Cloud Technologist, Oracle. He noted that SaaS Applications have become the predominant way enterprise applications are delivered. This shift is driven by promises of increased flexibility, access to cutting-edge technologies, and cost efficiencies. Learn about the key elements of a SaaS architecture that companies should be looking for and why, and how Oracle’s approach allows it to uptake the latest technologies and keep customer data secure.
“Accelerating Discovery with WEKA” was presented by Adam Fowler, Technical Product Marketing Manager, WEKA. Fowler said that the evolution of life sciences demands a data infrastructure capable of handling immense complex datasets to accelerate discovery and innovation. The WEKA Data Platform is a critical solution in this context, offering an advanced data management system tailored to meet the unique challenges of the life sciences sector. This session outlines the platform’s impact on accelerating discovery in life sciences through its unparalleled data processing capabilities, deployment flexibility, scalability, and performance.
“Artificial Intelligence: PrimateAI-3D and SpliceAI for Precision Medicine and Drug Discovery” was a talk delivered by Kyle Kai-How Farh, Vice President & Distinguished Scientist, Illumina. He said that PrimateAi-3D and Splice AI, leading algorithms from the Illumina Artificial Intelligence Laboratory, are disrupting how we approach drug discovery and precision medicine. In this session, we will review the data illustrating the impact that these two algorithms have on our understanding of variants of unknown significance (VUS) and the non-coding region of the genome – which can no longer be thought of as “junk.”
“Accelerating Medical Imaging Research and Innovation in the Biopharmaceutical Industry” was a presentation delivered by Mike Maker, Associate Director, Field Application Science, Flywheel. He noted that medical imaging has the power to inform the next medical breakthrough. Advanced imaging solutions and AI model development require a unified, agile, secure, and collaborative framework spanning the entire data lifecycle. Flywheel provides a cohesive cloud-based platform which accelerates the conversion of complex imaging data into analysis-ready datasets, while ensuring regulatory compliance through version control and provenance tracking — enhancing medical imaging research, boosting efficiencies, and spearheading trustworthy AI.
“Coding Against a Rare Pediatric Cancer: xCures & Slalom’s IT Mastery in the Moonshot” was a talk delivered by Mika Newton, CEO, xCures Inc., and Jeff Pierce, Managing Director, Slalom Inc. This talk details the work to create the DIPG-Onelink website and a health-data platform designed to help rare cancers as a part of President Biden’s Cancer Moonshot Initiative, which aims to reduce cancer death rates by 50% over the next 25 years. Learn about the website and the platform’s impact in the DIPG community and how they exemplify the tangible benefits of collaborative efforts in healthcare technology. (Editor’s note: DIPG stands for diffuse intrinsic pontine glioma).
“AI-Based Screening for Lung Cancer Patients” was delivered by Kamal Rawal, Professor and Head, Center for Computational Biology and Bioinformatics, Amity University (India). He noted that lung cancer stands as a significant global health challenge, with early diagnosis playing a pivotal role in improving patient outcomes. Current diagnostic methods, such as radiography, computed tomography (CT), positron emission tomography (PET), histopathology, and next-generation sequencing, are often time-consuming, costly, or require expert interpretation. These challenges are especially pronounced in resource-constrained countries like India, where widespread access to panel-based next-generation sequencing is limited. In addition, issues related to tissue adequacy in lung core biopsies and inherent intratumoral heterogeneity further emphasize the need for an AI-driven solution capable of autonomously detecting and learning nuanced lung nodule features from PET-CT and histopathological images while predicting the likelihood of mutations in critical oncogenes like EGFR.
In this study, we aimed to harness radiological and histopathological features from whole slide images (WSI) to predict lung malignancy, cancer type, and the presence of mutations, particularly in the EGFR gene. Building upon a prototype system, our project outlines the expansion of our initial work to create an end-to-end AI pipeline ready for deployment in oncology hospitals. This pipeline predicts essential lung nodule features, including radiological characteristics (e.g., malignancy, texture, margin), and histopathology (e.g., acinar, lepidic, papillary), and associates them with mutational information.
To date, we have assembled a valuable resource, comprising extracted lung cancer CT/PET scan datasets, alongside annotations from 2,000 Indian lung cancer patients, as well as an independent dataset sourced from the Cancer Image Archive and other repositories. We have also developed a user-friendly AI web tool tailored for clinical use, with preliminary tests demonstrating impressive accuracy (e.g., achieving a peak AUC of 91% in specific use cases). Our next phase anticipates the acquisition of an additional 2,000 images, along with annotations and mutational information from our collaborators. Our system is a robust machine learning (ML) platform, uniquely tailored for Indian lung cancer patients, incorporating multiple modules, including Region-Based Convolutional Neural Network (R-CNN) and other pertinent deep learning libraries. Our ultimate goal is to create a cost-effective AI platform that will empower oncologists, radiologists, and patients in resource-limited settings to achieve highly effective early lung cancer screening. This initiative promises to reduce costs and enhance treatment options, significantly benefiting patients.
“Multimodal Foundation Models for Biology: Learning the Language of Life” was presented by Zelda Mariet, Co-Founder and Principal Research Scientist, Bioptimus. Mariet said that foundation models (FMs) have catalyzed incredible AI breakthroughs in natural language. How far are we from achieving something similar for the biological language that connects DNA, proteins, cells, tissues, and organisms? In this talk, we unveil Bioptimus’s plan to bring together biological data to build a universal FM for biology—unleashing the full power of machine learning to transform our understanding of the living world, disease, and medicine.
“The UK Biobank Story: Transforming Future Health” was a presentation delivered by Naomi Allen, Professor, University of Oxford, Chief Scientist, UK Biobank. She said that the UK Biobank is a large-scale cohort study and biomedical database and is one of the world’s most important and widely-used health research resources, containing a wealth of data on genomics, lifestyle factors, and health outcomes on 500,000 participants. Its unique combination of scale, depth, maturity, and accessibility is far ahead of any comparable biological resources (and will remain so for many years to come). As a result, more than 30,000 researchers (10,000 in the US) from around the world are using UK Biobank, with exponential growth in publications and patents now emerging from it. 20 years from its initiation, UK Biobank has surpassed all expectations for its extraordinary impact on increasing understanding of the determinants of disease, identifying factors that can predict future disease risk more accurately, and identifying early disease biomarkers. During the next five years, UK Biobank would like to increase the depth of characterization of various types of risk factors for disease, and of the many different types of health outcome occurring in participants in order to substantially increase its value for discovery science.
“Finding Goldilocks—Can a Marriage of Quantum Mechanics and AI Unlock Successful Covalent Drug Development?” was presented by Johannes Hermann, Chief Technology Officer, Frontier Medicines, and Han Wool Yoon, AI Architect, Frontier Medicines. They noted that while some of the world’s most widely used medicines are covalent drugs (aspirin and penicillin, anyone?), most covalent medicines were discovered serendipitously and are often misunderstood. Covalent medicines form an irreversible bond and therein lie the benefits and the challenges, including finding just the right warhead that’s not too hot and not too cold. Breaking through the hype, can uniting AI and quantum mechanics unlock a successful future of deliberate exquisitely designed covalent medicines?
“Novel Machine Learning Approaches for Predicting the Functional Impact of Human Genetic Variants” was presented by Yuval Itan, Associate Professor, Icahn School of Medicine at Mount Sinai. He said that predicting the functional outcomes of human genetic variants poses significant challenges, primarily stemming from non-specific training relying on general pathogenicity or deleteriousness data. To tackle this issue, his group developed two novel genome-wide machine learning methods: (1) discerning whether a variant leads to gain-of-function or loss-of-function; and (2) predicting the disease phenotype associated with human genetic variants. Finally, he presented an efficient gene network clustering method designed for case-control gene associations within very small cohorts.
“Enabling a New Era of Precision Medicine for Complex Chronic Disease” was the title of a presentation by Simon Beaulah, Senior VP Healthcare & Head of US Operations, PrecisionLife. He said that the 20 years since the Human Genome Project have seen transformational advances in molecular understanding of cancers and rare disease, but precision medicine has had limited impact on improving treatment of complex, chronic indications in neuroscience, immune, and inflammatory disorders. To overcome this challenge, we have completely reimagined how to analyze, semantically integrate, and visualize patient and ‘omics data. This approach reveals deeper understanding of disease biology and enables development of diagnostic biomarkers and precision-targeted therapies for chronic diseases that affect billions of people.
“Addressing Antibody Property Prediction at Scale” was delivered by Thomas Blarre, Product Manager, Discngine. He noted that recent advances in high-throughput antibody discovery from Next Generation Sequencing and Phage Display allow for a larger number of candidates to test, proportionally increasing the potential late-stage failures. Therefore, identifying liabilities, developability issues, and immunogenicity early becomes essential. Here, we present an automatable, high-throughput, state-of-the-art, physics-based SaaS solution allowing to predict reliable 3D structures of various antibody types and derive 3D sensitive predictors for liabilities & developability.
END OF DAY
The day ended with a Best of Show Awards reception and poster viewing for in-person attendees. The third and final day of the conference will be Wednesday, April 17.