Each purified bait clone (DPiM_cDNA_bait_cDNA) was used to transiently transfect S2R+ cells in a 54ml culture. The individual S2R+ cultures were induced to produce FLAG-HA fusion protein by treatment with media containing 0.35 mM Cu2SO4 for 20 hours. Cells were lysed in buffer containing protease inhibitor and passed through a 0.45um durapore filter.

Each clarified lysate was bound overnight to crosslinked HA immunoaffinity resin. Unbound proteins were washed off with lysis buffer followed by PBS and then bound protein complexes were competitively eluted using synthetic HA peptide YPYDVPDYA (250 μg/ml) in PBS.

Copurified proteins were precipitated using trichloroacetic acid, washed with acetone, dried, digested overnight with trypsin, and analyzed by LC-MS/MS. LC-MS/MS spectral data were searched with SEQUEST (Eng et al., 2008) against a database of D. melanogaster proteins derived from FlyBase version 5.23. The LC-MS/MS identifications were filtered to, on average, a 0.3% peptide FDR.

The spectral data were searched with SEQUEST (Eng et al., 2008) against a database of D. melanogaster proteins derived from FlyBase version 5.23. The LC-MS/MS identifications were filtered to, on average, a 1.2% protein FDR and 0.3% peptide FDR. The compiled data set was filtered to a combined 0.8% FDR, and further post-processing was used to correct for column carryover issues. Both bait-prey and prey-prey protein interactions from coAP-MS data were analyzed and scored using HGSCore - a hypergeometric distribution error model, incorporating TSCs to improve the accuracy of co-occurrence prediction. A randomized data set of similar size was created to estimate FDR. Protein interactions were clustered using MCL (Enright et al., 2002).

The refined DPiM-2 dataset supercedes the original unfiltered protein interaction dataset, DPiM-1.