Dataset ------- **TOUGH-M2** comprises experimental target structures with binding pockets identified by [*e*FindSite 1.3][1]. The following files are included: ### Lists **`TOUGH-M2_target.list`** A list of 5873 targets **`TOUGH-M2_pocket.list`** A list of pockets identified by [*e*FindSite 1.3][1] * 1st column is the PDB-ID of target protein * 2nd column is the index of the best pocket * 3rd column is the accuracy of the best pocket (Matthews correlation coefficient over binding residues) **`TOUGH-M2_positive.list`** A list of 308665 positive pairs. These proteins bind chemically similar ligands despite having dissimilar global sequences and structures. * The first two columns are the PDB-IDs of target proteins * 3rd column is the sequence identity between the targets * 4th column is the structure similarity ([TM-score][2] by [Fr-TM-align][3]) * 5th column is the chemical similarity of bound ligands (Tanimoto coefficient by [kcombu][4]) **`TOUGH-M2_negative.list`** A list of 348926 negative pairs. These proteins have dissimilar global sequences and structures and bind chemically dissimilar ligands. * The first two columns are the PDB-IDs of target proteins * 3rd column is the sequence identity between the targets * 4th column is the structure similarity ([TM-score][2] by [Fr-TM-align][3]) * 5th column is the chemical similarity of bound ligands (Tanimoto coefficient by [kcombu][4]) ### Data **`TOUGH-M2_dataset.tar.gz`** A tarball containing the following files ([11asA][5] as an example): * target structure in PDB format (`11asA.pdb`) * target structure in PDBQT format (`11asA.pdbqt`) * bound ligand in PDB format (`11asA00.pdb`) * bound ligand in PDBQT format (`11asA00.pdbqt`) * bound ligand in SDF format (`11asA00.sdf`) * protein-ligand contacts by [LPC][6] (`11asA.lpc`) * sequence profile by [PROFILpro 1.1][7] (`11asA.profile`) * secondary structure by [PSIPRED 4.0][8] (`11asA.ss2`) * pockets by [*e*FindSite 1.3][1] (`11asA-efindsite.pockets.dat`) * template-target structure alignments by [*e*FindSite 1.3][1] (`11asA-efindsite.alignments.dat`) * ligands bound to templates extracted by [*e*FindSite 1.3][1] (`11asA-efindsite.ligands.sdf`) Results ------- Output files generated for the TOUGH-M2 dataset by several binding site alignment algorithms are available to facilitate comparative benchmarks. Pocket matching calculations were conducted against pockets predicted by [*e*FindSite 1.3][1]. The following programs are included: ### Template-free pocket alignment tools [APoc][9] **`APoc-TOUGH-M2_positive.score`** **`APoc-TOUGH-M2_negative.score`** PS-score values for positive and negative pairs. **`APoc-TOUGH-M2_positive.tar.gz`** **`APoc-TOUGH-M2_negative.tar.gz`** Output files from [APoc v1.0b15][9] for positive and negative pairs. [G-LoSA][10] **`G-LoSA-TOUGH-M2_positive.score`** **`G-LoSA-TOUGH-M2_negative.score`** GA-score values for positive and negative pairs. **`G-LoSA-TOUGH-M2_positive.tar.gz`** **`G-LoSA-TOUGH-M2_negative.tar.gz`** Output files from [G-LoSA v2.1][10] for positive and negative pairs. ### Template-based pocket alignment tools [*e*MatchSite][11] **`eMatchSite-TOUGH-M2_positive.score`** **`eMatchSite-TOUGH-M2_negative.score`** *e*MS-score values for positive and negative pairs. **`eMatchSite-TOUGH-M2_positive.tar.gz`** **`eMatchSite-TOUGH-M2_negative.tar.gz`** Output files from [*e*MatchSite 1.1][11] for positive and negative pairs. [1]: https://www.ncbi.nlm.nih.gov/pubmed/23838840 [2]: https://www.ncbi.nlm.nih.gov/pubmed/15476259 [3]: https://www.ncbi.nlm.nih.gov/pubmed/19077267 [4]: https://www.ncbi.nlm.nih.gov/pubmed/21736325 [5]: http://www.rcsb.org/pdb/explore.do?structureId=11AS [6]: https://www.ncbi.nlm.nih.gov/pubmed/10320401 [7]: https://www.ncbi.nlm.nih.gov/pubmed/15980571 [8]: https://www.ncbi.nlm.nih.gov/pubmed/10493868 [9]: https://www.ncbi.nlm.nih.gov/pubmed/23335017 [10]: https://www.ncbi.nlm.nih.gov/pubmed/26813336 [11]: https://www.ncbi.nlm.nih.gov/pubmed/25232727