Conference Agenda

We are witnessing an AI revolution. At the heart of this revolution are generative AI models that, powered by advanced architectures and large datasets, are transforming AI across a variety of disciplines. But how can AI facilitate and eventually enable groundbreaking discoveries in life sciences? How can it bring us closer to understanding biology – the functions of our cells, their alterations in diseases, and variations across species? In this talk, I will show how generative AI can uncover spatial relationships between cells, enabling the reassembly of tissues from dissociated single cells. Next, I will discuss the future of discovery in the era of generative AI and foundation models, highlighting the paradigm shift in machine learning required to revolutionize biology.

Rare diseases predominantly affect children, often causing premature death or lifelong disability. Over the past decade, the discovery of rare diseases has accelerated, driven by advances in genomic data generation and analysis, which now offer faster turnaround times at lower costs. This has led to a paradigm shift in the role of genomics in pediatric acute care. Despite these advancements, genome-dependent diagnostic rates remain low. As sequencing depth and speed have increased, the primary challenge has shifted from detecting genetic alterations to understanding their functional relevance. In this presentation, we will discuss novel deep learning methods to integrate multi-omics data—including whole-genome sequencing (WGS), RNA sequencing (RNA-seq), proteotyping, and metabolomics—to identify rare, deleterious genetic variants in children with life-threatening extreme phenotypes. We will also demonstrate the clinical impact of these methods by analyzing their effects at the gene, transcript, and proteome/metabolome levels.

The ability to generate multi-omic profiles at scale has created new opportunities to investigate the relationships across different omic read-outs. However, it comes with an increase in complexity making data analysis and interpretation more challenging. This heterogeneity of multimodal data pose significant analytical and interpretative challenges. Specifically, multi-modality creates correspondence problems, complicating the effective combination and analysis of diverse data types such as genomics, transcriptomics, proteomics data. Despite these challenges, leveraging synergies across datatypes may hold the potential to discover new insights by enabling a more comprehensive view of the underlying biological mechanisms. At single-cell resolution, the integration of multi-omics data allows us to leverage neural network architectures that can capture complex patterns and interactions within the data. In this talk, I will demonstrate examples of multi-comics integration and address correspondence problems demonstrating how we have leveraged the data to gain insights into biological mechanisms.