Construction and characterization of WI-2007. (a) Search spaces of WI-2007 and WI-2004 relative to the whole proteome. WI-2007 results from a 10,000 ORFs matrix screen, which is three times larger than WI-2004, corresponding to one fourth of the entire theoretical search space (screening completeness ~ 24%). (b) Pipeline used for WI-2007. ORFs from ORFeome v1.1 were transferred into DB and AD vectors by recombinational cloning, then transformed into yeast cells. Each bait was then mated with pools of 188 AD-ORFs. Two steps of phenotyping were performed to isolate positive colonies, which were used to PCR-amplify DB-ORFs and AD-ORFs for sequencing, leading to the identification of 5,822 Interaction Sequence Tags (ISTs). After exclusion of autoactivators and collapsing of all ISTs corresponding to the same, non oriented pair, an individual retest was performed to generate the final WI-2007 dataset. (c) WI-2007 characterization. Ten measurements were used (left to right) : proportions of cePRS-Y2H-v1, a random sample of WI-2007 and ceRRS-v1 observed in MAPPIT; number of interactions detected in the common search space of WI-2007 and WI-2004: in WI-2007, in both screens, and in WI-2004 AD-ORF; proportion of cePRS-v1 detected in an independent pairwise Y2H experiment; total number of interactions in WI-2007, proportion of ultra-conserved interologs and cePRS-v1 recovered in WI-2007. The Sampling errors on the 10 measurements are modeled with Beta distributions. Precision, sampling sensitivity, assay sensitivity and the total number of interactions in C. elegans are computed using a Monte-Carlo simulation. Y axis label (frequency) applies to all ten sampling distributions…

WI8 - an extended high-quality protein-protein interaction network. High-quality Y2H protein-protein interactions (PPI) from WI-2007 (1,816 interactions), WI-2004 (1,735 interactions), and diverse medium throughput biological processes based Y2H maps1,6–11 (554 interactions) were updated, filtered, assembled, and integrated into WI8. Two interactions, detected in both WI-2004 and WI-2007 with lower confidence were also “rescued”. The final dataset contains 3,864 high-confidence binary protein-protein interactions. In the network representation of WI8, the color of the edge indicates the dataset of origin: WI-2007 in red, WI-2004 in blue and biological process maps in green. Edges corresponding to several of these evidence types are shown in black, and edges corresponding to “rescued” interactions, i.e., supported by at least two lower confidence evidences, in grey. Only the main component of the network is shown.

Biological relevance. The Fold enrichment represents the frequency of functional linkage of protein pairs compared to random. The maximum number of genes associated to a particular GO term was used as an estimate of the functional specificity (the smaller the maximum number of genes per GO term, the higher the functional specificity), and the fold enrichments were calculated for several functional specificity levels (5, 20,100 and 400). (a) Enrichment of different components of WI8 and LCI binary for functional relationships. (b) Functional relationship enrichments for distinct types of experimental evidence. P-values assessing the difference between protein-protein interactions (PPI) and other types of evidence are shown for very specific GO terms (maximum 5 genes per GO term).

Examples of multiple evidence sub-networks. The networks represent genes-proteins (ellipses) relationships from several evidence sources: protein-protein interactions (PPI) from WI8 (separated according to origin): PPI from WI-2004 (dark blue), PPI from WI-2007 (red), PPI from biological processes maps (green), PPI from literature curation databases (magenta). The other types of functional links are: interologs (light blue), genetic interactions (violet), gross phenotypic similarity (light green), co-expression from a compendium of microarray experiments. Genes and their products are labeled using an unitalicized lower case version of the standard C. elegans three-letter system to reflect that the diagram combines links between both proteins and genes. (a) Genes related to RNA-binding. The genes annotated as ‘RNA-binding’ according to Wormbook are green ellipses in the network. White ellipses are genes-proteins linked to RNA-binding genes-proteins by at least one protein-protein interaction from WI8 and one additional piece of evidence. The inset shows the chronograms of sup-12 and mec-8 (left), and their predicted spatiotemporal pattern of interaction (right), corresponding to the stages and tissues where the interaction can take place. The color code of the chronograms represents the absolute GFP intensity measured (increasing values as black-green-yellow-white) for promoter(s) of the considered genes, along the worm length (x axis) and development stage (y axis)15. (b) Genes related to cell fusion. The genes-proteins annotated as ‘cell fusion’ according to Wormbook are green ellipses and genes-proteins linked to cell fusion genes-proteins by a protein-protein interaction from WI8 and one additional type of evidence are white ellipses.