Data-Driven Understanding of Human Disease: From Machine Learning Methods to Biological Discoveries

Explore data-driven approaches to understanding human disease in this Allen School Distinguished Lecture by Princeton's Olga Troyanskaya. Delve into the development and application of machine learning methods, including deep learning, Bayesian, and semi-supervised approaches, for biomedical data analysis. Discover how researchers tackle challenges in interpreting noncoding DNA, unraveling tissue-specific gene expression signals, mapping genetic circuits in disease-relevant cell types, and integrating biological knowledge from model organisms with human observations. Learn about innovative methods addressing these challenges and their applications to autism, Parkinson's, and cardiovascular disease. Gain insights into the genomic architecture of disease, the complexities of DNA sequences in different organs, and the use of model organisms in human disease research. Follow Troyanskaya's journey through computational biology, bridging computer science and molecular biology to develop better methods for analyzing diverse genomic data and modeling protein function and interactions in biological pathways.

Introduction
The main problem
The human genome
One fundamental puzzle
How does a single letter change
Complex networks
Model organisms
How we started answering the first question
What happens if we are outside of the genes
Deep convolutional neural net
Neanderthal genome
Does it actually work
Humanbase
Cross organism networks
Naive Bayesian classifier
Semisupervised training
Free time
How do we study these mechanisms
Can we systematically integrate model organism information with the human quantitative genetic studies
Parkinsons disease
Top genes
Bodybuilding
Keynote
Proteinprotein interactions
Dynamics with networks
Suppressor screens
Subsampling