RT Journal Article
SR Electronic
T1 Unsupervised extraction of stable expression signatures from public compendia with eADAGE
JF bioRxiv
FD Cold Spring Harbor Laboratory
SP 078659
DO 10.1101/078659
A1 Jie Tan
A1 Georgia Doing
A1 Kimberley A. Lewis
A1 Courtney E. Price
A1 Kathleen M. Chen
A1 Kyle C. Cady
A1 Barret Perchuk
A1 Michael T. Laub
A1 Deborah A. Hogan
A1 Casey S. Greene
YR 2017
UL http://biorxiv.org/content/early/2017/04/10/078659.abstract
AB Cross experiment comparisons in public data compendia are challenged by unmatched conditions and technical noise. The ADAGE method, which performs unsupervised integration with neural networks, can effectively identify biological patterns, but because ADAGE models, like many neural networks, are over-parameterized, different ADAGE models perform equally well. To enhance model robustness and better build signatures consistent with biological pathways, we developed an ensemble ADAGE (eADAGE) that integrated stable signatures across models. We applied eADAGE to a Pseudomonas aeruginosa compendium containing experiments performed in 78 media. eADAGE revealed a phosphate starvation response controlled by PhoB. While we expected PhoB activity in limiting phosphate conditions, our analyses found PhoB activity in other media with moderate phosphate and predicted that a second stimulus provided by the sensor kinase, KinB, is required for PhoB activation in this setting. We validated this relationship using both targeted and unbiased genetic approaches. eADAGE, which captures stable biological patterns, enables cross-experiment comparisons that can highlight measured but undiscovered relationships.