Recent advances in disease biology, combination therapies, and treatment modalities have positively impacted human health.  Paradoxically, these advances created a complex and fragmented clinical landscape, amplifying the need for precision medicine approaches to enable drug development. In this talk, I will present several case studies of contemporary biomarker discovery efforts that leverage diverse data types (-omics, pathology, imaging, real world data) and modern computational methods to advance drug development programs.

To facilitate molecular studies of Alzheimer's disease (AD), we have developed Alzheimer DataLENS, a web-based portal that enables researchers to share, browse, and visualize comprehensive results from bioinformatics analysis of public omics datasets. Currently, the portal has data from over 50 genetic, proteomics, and transcriptomics studies. Recently, several studies have used single-nucleus transcriptomics to characterize the brains of patients with Alzheimer's disease. We leveraged these datasets to develop innovative methods of visualizing gene expression across various cell types, brain regions, and disease states.

To meet the growing demands from scientists to effectively extract deep insights from single cell RNA sequencing, spatial transcriptomics, and emerging multiome datasets, we developed cellxgene VIP (Visualization In Plugin), a front end interactive visualization plugin of cellxgene framework, which greatly expanded capabilities of the base tool in the following aspects. First, it generates a comprehensive set of eighteen commonly used quality control and analytical plots in high resolution with highly customizable settings in real time. Second, it provides more advanced analytical functions to gain insights on cellular compositions and deep biology, such as marker gene identification, differential gene expression analysis, and gene set enrichment. Third, it empowers advanced users to perform analysis in a Jupyter notebook like Command Line Interface (CLI) environment by programming in Python and/or R directly without containing to available interactive functional modules. Finally, it pioneers methods to visualize multi-modal data, such as spatial transcriptomics embedding aligned with histological image on one slice or multiple slices in a grid format, and the latest 10x Genomic Multiome dataset where both DNA accessibility and gene expression in the same cells are measured, under the same framework in an integrative way to fully leverage the aforementioned functionalities. Taken together, the open-source tool makes large scale single cell data visualization and analysis more accessible to biologists in a user-friendly manner and fosters computational reproducibility by simplifying data and code reuse through the CLI.  Going forward, it has the potential to become a crowdsourcing ecosystem for the scientific community to contribute even more modules to the Swiss Army knife of single cell data exploration tool.

Abstract to be announced.

Production bioinformatics at Sema4 can be thought of as data ops - a peer to the lab ops organization. We operate 24/7 to deliver correct and timely results on NGS and other data for thousands of samples per week. Chris Dwan, Senior Vice President will introduce the Prod BI organization and systems architecture, with a focus on what it takes to run bioinformatics in production rather than for R&D or pure research. 

While whole genome sequencing in UK Biobank and other cohorts delivers gene-phenotype associations at unprecedented scale, further analysis is required to validate these associations with evidence for biomarker or drug target identification. In this talk, we describe collaborations forged in the J&J innovation eco-system with partners who have developed experimental and/or informatics approaches capable of elucidating mechanisms and delivering actionable target or biomarker insights and opportunities.

Translational research requires pairing a variety of data modalities with the appropriate cyber-infrastructure and computational platforms to facilitate and expedite analysis. We introduce the UK Biobank - a large-scale biomedical database containing in-depth genetic and health information from half a million UK participants. The database, which is regularly augmented is globally accessible to approved researchers undertaking research into life-threatening diseases. We outline how the data can be accessed and analyzed via a fit-for-purpose cloud-based analysis platform as well as highlight impactful uses cases.

Accurately determining structural variants from WGS data is a non-trivial problem especially from large population-based cohorts such as that generated from 500K individuals within the UK Biobank via a public-industry consortium. We highlight an example of combining state-of-art algorithm for structural variation calling developed by Illumina with massive geographically distributed cloud-based computational resources provided by AWS. We anticipate that results of this analysis will be publicly available to the global scientific community in the near future.