Cover Picture: Proteomics 6'10

Looking into an uncrystallized ball For want of reliable and reproducible methods of crystallizing membrane proteins, it has been difficult to make predictions of their three‐dimensional structures for drug research purposes. Taking a calculated approach, Fagerberg et al . applied seven protein structure modeling methods to >21 000 human gene sequences. The search returned 5500–7600 proteins with predicted alpha helical transmembrane segments. Using a consensus method, they estimated that 5500 (26%) of human genes code for membrane proteins with between one and seven transmembrane segments. (GPCRs show seven segments.) A word of warning from the authors – never take just one method's result as gospel. Fagerberg, L. et al., Proteomics 2010, 10, 1141–1149. Re‐sampling statistics for sensitivity An increasingly common means of label‐free relative quantitation of protein is spectral counts (SC). The mass spectroscopy method yields linear results although there has been no rigorous statistical foundation established. Here, Little et al. tame the non‐normal distribution by applying reSASC – resampling‐based Significance Analysis for Spectral Counts. Similar data were pooled and then re‐sampled to create synthetic sets of SCs that were well behaved. The method was tested against two published data sets – yeast and cystic fibrosis. The new test was compared with conventional t ‐tests and the Wilcoxon test as well as the SC‐oriented Qspec and Spectral Index tests. The conventional SC methods were able to detect a twofold change in expression, whereas reSASC was able to detect 1.7‐fold in cystic fibrosis and 1.2‐fold changes in yeast GAPDH. Little, K. M. et al., Proteomics 2010, 10, 1212–1222. Have a heart, but not too big a heart Over 30% of the US deaths were due to heart diseases in 2005. Many of these deaths could be prevented if the symptoms of heart disease were recognized early enough. The challenge is finding a “molecular signature” for the treatable early stage disease. A promising approach is proteomic analysis using shotgun ESI‐MS/MS technology. Isserlin et al. turned up over 7000 differentially expressed proteins when it was applied to a three time‐point mouse model of dilated cardiomyopathy. After normalization for size, this abundance of non‐Gaussian data was dealt with using Gene Set Enrichment Analysis and a new technique they developed, Enrichment Mapping, to do over‐representation analysis to find functionally related gene sets. The procedure is very efficient at grouping even weakly expressed genes on the basis of commonality in annotation. Reanalysis suggests that up‐regulation of ATP generation (NADH dehydrogenase, ATP synthase) is an early marker for dilated cardiomyopathy. Isserlin, R. et al. Proteomics 2010, 10, 1316–1327.