Unsupervised Extraction of Stable Expression Signatures from Public Compendia with an Ensemble of Neural Networks.

Cross-experiment comparisons in public data compendia are challenged by unmatched conditions and technical noise. The ADAGE method, which performs unsupervised integration with denoising autoencoder neural networks, can 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.

Evolving new protein-protein interaction specificity through promiscuous intermediates.

Interacting proteins typically coevolve, and the identification of coevolving amino acids can pinpoint residues required for interaction specificity. This approach often assumes that an interface-disrupting mutation in one protein drives selection of a compensatory mutation in its partner during evolution. However, this model requires a non-functional intermediate state prior to the compensatory change.

A DNA damage-induced, SOS-independent checkpoint regulates cell division in Caulobacter crescentus.

Cells must coordinate DNA replication with cell division, especially during episodes of DNA damage. The paradigm for cell division control following DNA damage in bacteria involves the SOS response where cleavage of the transcriptional repressor LexA induces a division inhibitor. However, in Caulobacter crescentus, cells lacking the primary SOS-regulated inhibitor, sidA, can often still delay division post-damage.

Regulated proteolysis of a transcription factor complex is critical to cell cycle progression in Caulobacter crescentus.

Cell cycle transitions are often triggered by the proteolysis of key regulatory proteins. In Caulobacter crescentus, the G1-S transition involves the degradation of an essential DNA-binding response regulator, CtrA, by the ClpXP protease. Here, we show that another critical cell cycle regulator, SciP, is also degraded during the G1-S transition, but by the Lon protease.

Spatial tethering of kinases to their substrates relaxes evolutionary constraints on specificity.

Signal transduction proteins are often multi-domain proteins that arose through the fusion of previously independent proteins. How such a change in the spatial arrangement of proteins impacts their evolution and the selective pressures acting on individual residues is largely unknown. We explored this problem in the context of bacterial two-component signalling pathways, which typically involve a sensor histidine kinase that specifically phosphorylates a single cognate response regulator.