This presentation will address the problem of biological complexity in which many factors, each of small effect size, collectively influence disease risk, development, complexity, and response to therapy in cancer and other complex diseases. By using innovative computational methods built around network representations of biological interactions, we can gain insight into the disease process, develop predictive biomarkers, and identify possible avenues of therapeutic intervention.

Synlogic is delivering the power of synthetic biology to medicine, combining the precision of engineering with rational drug development. With a premiere synthetic biology platform, we design Synthetic Biotic medicines to treat disease in new ways. Synthetic biology brings engineering principles to science enabling the design of new biological systems, genetic circuits and molecular components that can create medicines that can have a potentially meaningful impact for patients.

Quality bioinformatics pipelines are indispensable for getting insights out of molecular data, potentially leading to breakthroughs in drug development. However, it is difficult to identify the best tools for these pipelines. Recently launched Algorithm Benchmarking Consortium is an attempt at a pre-competitive effort to systematically evaluate bioinformatics tools. This talk will focus on how the consortium was conceived, how it operates, and what insights it brings to the members.

There have been many recent advances in algorithms and tools for analyzing cancer genomes and making them useful in a research and clinical context. I will discuss how these developments are improving care including the use of a new advanced genome reference.

Long-read sequencing became widely adopted due to its improvements for Structural Variation detection and methylation calling. But, it is hard to fully utilize the generated data and thus many studies focus either on SV, methylation, or phasing of SNVs. We developed PRINCESS, PRINCESS starts with a raw sequence and provides within hours a fully phased SNV, SV, and methylation call set. This allows a comprehensive study of the sample at scale and for a wide number of scientists. PRINCESS achieves high accuracy (SNVs: 96.21%-97.92%, SVs: 85.60%-90.91%) and long phasing 821,907 bp (N50).

Machine learning and deep learning models are rapidly being adopted throughout the entire drug discovery value chain. Deploying them productively in business processes is a critical step towards generating business value and transforming pharma research. Handling these productive deployments and managing them throughout their life cycle requires dedicated and collaborative efforts across the organization.

Systems that execute enterprise level automated data transfer are often black boxes with limited configurability and access. This leaves R&D organizations reliant on vendor services for even simple workflow changes. To address this, we introduce a framework that allows for the admin-level user to configure and design an analytical data marshalling system. This framework can incorporate source monitoring, data transfer, data processing, databasing, and measure system performance metrics.

This talk will help you understand some of the opportunities to use open and free-to-use chemical data available from resources such as PubChem, which I manage at the National Center for Biotechnology Information (NCBI).  By the end of this talk, you should have some sense of the ways you can access open chemical information content from within your workflows and how it might help you or your colleagues improve the science that you do.

We developed a supervised machine learning pipeline using longitudinal Electronic Health Records (EHR) to accurately predict COVID-19 related health outcomes. In particular, we developed unique and effective data processing algorithms, then trained models using state-of-the-art machine learning strategies combined with different parameter settings and feature selection. In addition, we identified top COVID-19 risk factors, which are consistent with epidemiologic findings. This pipeline achieved top performance at VHA/precisionFDA COVID-19 modeling challenge.