Representational Learning in Genome Biology and Medicine
Biomedical and translational science researchers are increasingly accumulating genomics data that complement clinical and imaging data used in clinical records. As a popular subfield of machine learning, representational learning (RL) may help drug developers understand the complex relationships among human biology, genetic factors, and therapeutic agents by learning a representation of the data. RL aims to capture the underlying structure and variations in the data to make accurate predictions or inferences about new gene-drug-disease data. It can cover a broad spectrum of unsupervised, deep, and transfer learning on vast biological data sets beyond the medical image.
This session aims to bring bioinformaticians, data scientists, and biomedical domain experts together in the GLBIO community to rethink how to leverage AI/ML power for a better disease biology understanding, which aids precision medicine and tailored therapeutic development. This session will draw a broad spectrum of trainees from bioinformatics, AI/ML, biology, pharmacology, clinical practice, and pharmaceutical development. In this session, we will solicit late-breaking primary research on these topics and lead the discussions among experts for the following topics:
• How can we characterize and reduce data noises from different omics platforms for AI/ML applications?
• What are the data pre-processing techniques, including statistical, computational, AI/ML and mathematical models, to help normalize, combine, and compare multi-omics results from different experimental conditions?
• How can we balance knowledge-guided and knowledge-agnostic approaches in the development of AI/ML predictive model?
• How can we infer the cellular states and their dynamics, including the changing molecular profiles, networks, and pathways, which goes beyond the AI/ML blackbox models?
• How can we infer multi-scale causal network knowledge using AI/ML?
- Jake Y. Chen, University of Alabama at Birmingham
URL: Coming Soon
Bioinformatics for the people, by the people
In the information age, bioinformatics has become a core component of the biological
and biomedical research pipeline. The development of tools and infrastructure to
facilitate access of this technology to the entire scientific community is key to enable
collaboration of large multidisciplinary research teams and accelerate the discovery
The discoveries and applications enabled by bioinformatics methods have a profound
impact on our understanding of living systems and practice of medicine. To ensure a
broad adoption by our societies of these scientific advances, it is also essential to
communicate to the public the motivation and foundation of this research.
Simultaneously, the deployment of platforms to bring bioinformatics to the public opens
new opportunities to accelerate biological and biomedical research.
Crowdsourcing, or the practice of obtaining services by soliciting contributions from a
large group of individuals through the internet, is now a well-established practice to
conduct projects requiring humans to perform a large number of repetitive tasks that
could hardly be completed by computers at the same level of accuracy and confidence.
We refer to online citizen science when this technology is used to gather crowds of
volunteers through the web for addressing scientific challenges.
The application of this technology to biological problems has already contributed to
major scientific advances (e.g. prediction and design of molecules, mapping of neurons)
and had a positive impact on the public perception and understanding of bioinformatics
This special session will introduce the GLBIO audience to the motivations, principles
and opportunities offered by the democratization of bioinformatics technologies.
Importantly, we will emphasize the methodological contributions and describe upcoming
- Jérôme Waldispühl, McGill University
URL: Coming Soon