Molecular chaperones play an important role in protein homeostasis and the cellular response to stress. In particular, the HSP70 chaperones in yeast mediate a large volume of protein folding through transient associations with their substrates. This chaperone interaction network can be disturbed by various perturbations, such as environmental stress or a gene deletion. Here, we consider deletions of two major chaperone proteins, SSA1 and SSB1, from the chaperone network in Sacchromyces cerevisiae. We employ a SILAC-based approach to examine changes in global and local protein abundance and rationalise our results via network analysis and graph theoretical approaches. Although the deletions result in an overall increase in intracellular protein content, correlated with an increase in cell size, this is not matched by substantial changes in individual protein concentrations. Despite the phenotypic robustness to deletion of these major hub proteins, it cannot be simply explained by the presence of paralogues. Instead, network analysis and a theoretical consideration of folding workload suggest that the robustness to perturbation is a product of the overall network structure. This highlights how quantitative proteomics and systems modelling can be used to rationalise emergent network properties, and how the HSP70 system can accommodate the loss of major hubs.

Mass spectrometry (MS)-based quantitative proteomics experiments typically assay a subset of up to 60% of the ≈20 000 human protein coding genes. Computational methods for imputing the missing values using RNA expression data usually allow only for imputations of proteins measured in at least some of the samples. In silico methods for comprehensively estimating abundances across all proteins are still missing. Here, a novel method is proposed using deep learning to extrapolate the observed protein expression values in label-free MS experiments to all proteins, leveraging gene functional annotations and RNA measurements as key predictive attributes. This method is tested on four datasets, including human cell lines and human and mouse tissues. This method predicts the protein expression values with average R2 scores between 0.46 and 0.54, which is significantly better than predictions based on correlations using the RNA expression data alone. Moreover, it is demonstrated that the derived models can be "transferred" across experiments and species. For instance, the model derived from human tissues gave a R2=0.51 when applied to mouse tissue data. It is concluded that protein abundances generated in label-free MS experiments can be computationally predicted using functional annotated attributes and can be used to highlight aberrant protein abundance values.