“CompBio V2.5 – New Functionality, Enhanced Memory Model, and Comparative Analytics”

Hosted by the McDonnell Genome Institute (MGI)

CompBio(version 2.5), a next-generation omics analysis platform developed at Washington University, is now in production and available to all Washington University investigators.  With utilization by ~200 users/labs at Wash U and cited research application in high-impact journals such as Nature, Cell, PNAS, Cell Metabolism, Journal of the American College of Cardiology, and numerous others, we continue to receive requests for enhancements to key functionality within the platform.  

The seminar will cover the addition of several highly requested features in v2.5 including a sophisticated automated annotation system.  The new auto-annotation feature can be used on both new and older projects and provides annotation at the theme and project levels.  Unlike ontology-based tools such IPA and GSEA, the CompBio auto-annotation algorithm annotates themes not only based on their own content but that of their neighbors as well.  As such, the biological processes and pathways have a higher degree of contextual accuracy.  Additional functionality discussed will include features to better enable generation of publication quality figures and enhancements to the PCMM (core memory model) system to identify newer biological concepts. Finally, we will show some new comparative analytics with commonly used pathway tools.  A new user manual will also be available with the v2.5 release.

About CompBio:  CompBio is a next-generation Artificial Intelligence platform for biological discovery that combines a novel, human-like Contextual Memory Model (PCMM) system with advanced generalized signal detection and machine learning algorithms that can compute across a comprehensive representation of biological knowledge.  The current PCMM is built on the May 2022 release of PubMed and contains 33 million+ abstracts and 3 million+ full text articles.  Users input a biological entity list of interest and CompBio will build a knowledge map of the biological processes and pathways associated with those entities in a contextually relevant manner that is not ontology dependent.  Biological entities currently understood by CompBio include gene/protein names, metabolites, microbes (genus & species), and microRNAs.  Additionally, complex cross-project analyses can be performed using the Assertion Engine tool within the CompBio suite.  The Assertion Engine can identify contextually preserved biology between related projects such as human disease and animal models, drug treatments on a common background, time series responses, and numerous other comparators.  CompBio is a web-based application and can be utilized freely by all Wash U investigators and their laboratories. 

For inquiries contact the McDonnell Genome Institute.