This dataset is made available on dryad via the eLife journal as part of the following publication: Sahoo A, Khare S, Devanarayanan S, Jain P, Varadarajan R. Residue proximity information and protein model discrimination using saturation-suppressor mutagenesis. Elife. 2015 Dec 30;4. pii: e09532. doi: 10.7554/eLife.09532. [Epub ahead of print] PubMed PMID: 26716404. In the above mentioned publication, the CcdB decoy dataset has been employed to probe the utility of the experimentally obtained contact information in model discrimination. The models were scored based on ContactScore (Csc), which was defined as the number of times the experimentally identified residue contact pairs (6 pairs) are within a cutoff distance of 7Å of each other in a given model. CSc is an integral value ranging from 0 to 6 since there are six (PIM, proximal suppressor) pairs in case of CcdB. From the CcdB decoy dataset, CSc was able to recover 100% of models with backb- -one RMSD <2Å, 66% of mdoels with backbone RMSD <4Å and subsequently plateauing to 0% for models >5Å, reflecting the sensitivity and accuracy of the parameter. ########## Background Information ########## This dataset was generated as part of the following publication: Adkar BV, Tripathi A, Sahoo A, Bajaj K, Goswami D, Chakrabarti P, Swarnkar MK, Gokhale RS, Varadarajan R. Protein model discrimination using mutational sensitivity derived from deep sequencing. Structure. 2012 Feb 8;20(2):371-81. doi: 10.1016/j.str.2011.11.021. PubMed PMID: 22325784. We have derived an empirical parameter, 'RankScore' from sequencing analysis of a single-site saturation mutagenesis library of Controller of Cell Division or Death B (CcdB) protein. RankScore was observed to correlate (r=0.6) with residue depth (Chakravarty and Varadarajan, 1999) derived from the crystal structure of CcdB (PDB id 3VUB). If a structural model is similar to the native structure, residue depth from the native structure and models should be highly correlated. Since we have shown that RankScore is correlated to residue depth from the na- -tive structure, in principle, the correlation coefficient between RankScores and model residue depths can be used for model discrimination. To test this hypothesis, a decoy set was generated for CcdB by refining models obtained from threading. 27 threading alignments (from eight different dimeric templates with different RMSD values from the CcdB crystal structure, PDB id 3VUB; Loris et al., 1999) were used for model generation. Each of these models was further refined using the Rosetta refinement protocol (Raman et al., 2009) to generate 300-500 models per alignment (see Experimental Procedures Adkar et al., 2012). ########## Dataset Generation ########## The query protein sequence was submitted to the LOMETS metaserver (Wu and Zhang, 2007). The search database contained protein sequences with <30% identity to the query sequence. The models were generated using alignments only with dimeric te- -mplates. 7-9 templates were selected from various backbone RMSD ranges with respect to the crystal structure of the target (care was taken not to overpopulate any sp- -ecific RMSD range) and further refined using Rosetta with the following options: rosetta -relax -farlx -ex1 -ex2 -short_range_hb_weight 0.5 -long_range_hb_weight 1.0 -farlx_cycle_ratio 1.0 to generate ~300-500 models for each template. Dimeric models of CcdB generated by threading were refined using the following opti- -ons of Rosetta (version 3.2): rosetta -relax: thorough -relax:chi_move -symmetry:symmetry_definition. The backbone RMSD was calculated with respect to the native crystal structure using the g_confrms tool of Gromacs (version 4.0.5; Hess et al., 2008; Van Der Spoel et al., 2005). ########## Dataset Structure ########## The 27 different folders correspond to the above mentioned 27 threading alignments. Each of these 27 folders, contains a folder named 'PDBs'. All the model PDB files are stored in the 'PDBs' folder. A list of all the 10,659 models is also provided in a file named 'list'. ########## Contact Information ########## Please write to varadar@mbu.iisc.ernet.in in case of any additional queries. ########## Thank you! ##########