BM5-clean

Docking benchmark 5.5 (BM5 and BM5.5) - cleaned and ready to use for HADDOCK

This is the docking and binding affinity benchmark described in:

T Vreven, I.H. Moal, A. Vangone, B.G. Pierce, P.L. Kastritis, M. Torchala, R. Chaleil, B. Jimenez-Garcia, P.A. Bates, Juan Fernandez-Recio, A.M.J.J. Bonvin and Z. Weng. Updates to the integrated protein-protein interaction benchmarks: Docking benchmark version 5 and affinity benchmark version 2.
J. Mol. Biol. 19, 3031-3041 (2015). https://doi.org/doi:10.1016/j.jmb.2015.07.016

and its antibody-antigen extension, Docking Benchmark 5.5 (https://zlab.wenglab.org/benchmark/):

J.D. Guest, T. Vreven, J. Zhou, I. Moal, J.R. Jeliazkov, J.J. Gray, Z. Weng and B.G. Pierce. An expanded benchmark for antibody-antigen docking and affinity prediction reveals insights into antibody recognition determinants.
Structure 29, 606-621 (2021). https://doi.org/10.1016/j.str.2021.01.005

HADDOCK-ready contains 271 complexes: the 230 BM5 entries and the 41 entries added in BM5.5 (antibody-antigen and nanobody-antigen complexes).

The repository contains the following information:

HADDOCK-ready

The directory contains HADDOCK-ready files for each entry of BM5 and BM5.5. Each sub-directory (one per complex) contains the following files:

• XXX_r_u.pdb : Unbound receptor PDB
• XXX_l_u.pdb : Unbound ligand PDB
• XXX_r_b-matched.pdb : Matched bound receptor PDB
• XXX_l_b-matched.pdb : Matched bound ligand PDB
• XXX_target.pdb : the reference, matched bound complex (receptor chain A, ligand chain B)

And for each PDB there is an associated .info file providing statistics of the PDB content.

The bound and unbound forms of each partner have the same residues, with the same numbering and residue types (receptor chain A, ligand chain B, numbered sequentially without insertion codes):

• Selenomethionines are MET (ATOM records, SD, element S); other modified residues are given as their parent residue (e.g. methyl-His as HIS, Cys sulfinic acid as CYS).
• Residues present in both the unbound and the bound PDB entries are included, also when they have alternate locations (conformation A is used) or insertion codes. Residues missing from either PDB entry are left out, as are residues of different type in the bound and unbound forms.
• TER records separate different chains (e.g. when two chains are merged into one) and co-factors, not chain breaks.
• Co-factors and ligands present in the unbound structures are kept as HETATM.

Two distance restraints files are present:

• XXX_ambig.tbl : Ambiguous interaction restraints for the true interface of the reference complex measured with a 3.9A cutoff, generated with haddock-restraints ti XXX_target.pdb 3.9 (on the reference without HETATM)
• XXX_ambig5.tbl: Same with a 5.0A cutoff (haddock-restraints ti XXX_target.pdb 5)
• XXX_hbonds.tbl : If present, unambiguous restraints keeping together the disconnected parts of each unbound partner (chain breaks, merged chains and co-factors), generated with haddock-restraints restraint on XXX_r_u.pdb and XXX_l_u.pdb and concatenated. Absent when neither partner has chain breaks or co-factors.

And if there is a co-factor or ligand in the structure:

• XXX_ligand.param : the ligand parameter file as generated by PRODRG
• XXX_ligand.top : the ligand topology file as generated by PRODRG

Further each sub-directory contains an ana_scripts directory with the reference files for the analysis:

• target-unbound.pdb : the unbound complex (XXX_r_u.pdb and XXX_l_u.pdb; each unbound partner is superimposed onto its bound partner in the reference complex over its interface, backbone of target.izoneA or target.izoneB)
• target.contacts3.9 : intermolecular contacts at 3.9A cutoff
• target.contacts5 : intermolecular contacts at 5A cutoff used for calculating the fraction of native contacts
• target.izone : the interface definition for i-RMSD calculations with ProFit (derived from all residue contacts at 10A)
• target.izoneA : same as target.izone but for chainA only
• target.izoneB : same as target.izone but for chainB only
• target.lzone : the zone definition for l-RMSD calculations with ProFit

For complexes with a symmetric partner (e.g. a homo-oligomer offering several equivalent binding sites), ana_scripts_2 (and ana_scripts_3, ...) contain the same files for the symmetry-related reference complexes, in which one partner is placed on another binding site by a symmetry operation of the other partner, so that models can be evaluated against each equivalent site:

• target.pdb : the symmetry-related reference complex
• target.contacts5, target.izone, target.izoneA, target.izoneB, target.lzone : as above

The HADDOCK-ready/ana_scripts directory contains the analysis scripts:

• i-rmsd_to_xray.csh: csh script to calculate i-RMSDs with ProFit from HADDOCK file.nam files
• l-rmsd_to_xray.csh: csh script to calculate l-RMSDs with ProFit from HADDOCK file.nam files
• fraction-native.csh : csh script to calculate the fraction of native contacts from HADDOCK file.nam files
• cluster-fnat.csh : a script that generate cluster stats including RMSD and Fnat values
• run_all.csh : a csh script that runs the complete analysis of all three stages of HADDOCK

Note that the paths in the various analysis scripts must be adapted to your directory structure. This can be done by running the HADDOCK-ready/scripts/setup-ana_scripts.csh script with as argument the directory name of all entries

The categories directory lists the complexes by type (one complex per line), with the complex category labels of the Docking Benchmark 5.5:

• EI.txt : Enzyme-Inhibitor
• ES.txt : Enzyme-Substrate
• ER.txt : Enzyme complex with a regulatory or accessory chain
• AA.txt : Antibody-Antigen
• AS.txt : Antigen - Single domain Antibody
• OG.txt : Others, G-protein containing
• OR.txt : Others, Receptor containing
• OX.txt : Others, miscellaneous
• AB.txt : Antibody-Antigen complexes without an unbound antibody structure (BM5 entries, removed from BM5.5)
• non-antibody.txt : all complexes except the antibody and nanobody ones (all but AA, AS and AB)
• antibody-nanobody.txt : antibody and nanobody complexes (AA and AS)

Finally, the HADDOCK-ready directory also contains pre-calculated i-RMSD values for the superimposed unbound structures onto the reference complex and for each separate interface (ProFit, backbone atoms N, CA, C and O over the target.izone, target.izoneA and target.izoneB zones):

• i-RMSD.dat : Interface RMSD unbound superimposed versus reference, sorted in the order of the directory listing
• i-RMSD-sorted.dat : Interface RMSD unbound superimposed versus reference, sorted from small to large
• i-RMSD_r.dat : Interface RMSD of the unbound receptor interface versus reference, sorted in the order of the directory listing
• i-RMSD_r-sorted.dat : Interface RMSD of the unbound receptor interface superimposed versus reference, sorted from small to large
• i-RMSD_l.dat : Interface RMSD of the unbound ligand interface versus reference, sorted in the order of the directory listing
• i-RMSD_l-sorted.dat : Interface RMSD of the unbound ligand interface superimposed versus reference, sorted from small to large

Other sub-directories:

• scripts : setup-ana_scripts.csh, to adapt the paths in the analysis scripts

haddock3-scripts

Configuration files to run the benchmark with haddock3 using haddock-runner:

• unbound-bench.yaml : benchmark configuration with two scenarios starting from the unbound structures, all-atom (all-atom_unbound) and coarse-grained Martini 2 (coarse-grain_unbound), using the XXX_ambig5.tbl restraints and evaluating against XXX_target.pdb
• input.list : the list of unbound receptor and ligand PDB files of all complexes except the AB ones (antibody-antigen complexes without an unbound antibody, HADDOCK-ready/categories/AB.txt)
• input-<NAME>.list : the same for the complexes of each list in HADDOCK-ready/categories (e.g. input-EI.list, input-antibody-nanobody.list, input-non-antibody.list)
• haddock3.sh : wrapper script calling haddock3 (paths must be adapted to your installation)

scripts

• scripts regenerating the HADDOCK-ready restraint and analysis files (contacts, ProFit zones, XXX_ambig.tbl, XXX_ambig5.tbl, XXX_hbonds.tbl, symmetry-related references, i-RMSD tables);
• curation: the scripts used to check the structures against the PDB entries and correct them.

See the README in that directory.

Changes

The 41 BM5.5 entries (1S78, 2DD8, ..., 6OC3) were added from the BM5.5 distribution (benchmark5.5.tgz) with scripts/curation/add_bm55.py: bound and unbound chains paired, only residues present in both forms with the same type kept, insertion codes removed and chains merged into a single chain per partner (later chains shifted by multiples of 500). Symmetry-related references (ana_scripts_K) were created for the new entries with a symmetric partner (2FJG, 3WD5, 5HYS, 5KOV, 5WUX, 5Y9J, 6EY6) with scripts/make-symmetry-refs.py.

All unbound partners (BM5 and BM5.5) were superimposed onto their bound partners over their interface (scripts/superimpose-unbound.py); the complex i-RMSD values now agree with those of the BM5.5 table (median difference 0.03 A).

The HADDOCK-ready files were checked against the PDB entries and corrected:

• Bound numbering matched to the unbound numbering where it differed (1OPH ligand, 1FAK and 1AZS receptors; the protomers 2 and 3 of the 4FQI ligand trimer were swapped in the bound form).
• Residues present in both the bound and unbound PDB entries but missing from both files were added: selenomethionines (as MET), other modified residues (as their parent residue), and residues with alternate locations or insertion codes. Remaining selenium atoms in MET were converted to sulfur.
• TER records at chain breaks were removed.
• target-unbound.pdb files were rebuilt from the current unbound files.
• The symmetry-related references (ana_scripts_K) were rebuilt from the current reference complexes with exact symmetry operations (except 1QFW, whose two hCG subunits are only pseudo-symmetric).
• Contacts, ProFit zones, XXX_ambig.tbl, XXX_ambig5.tbl, XXX_hbonds.tbl and the i-RMSD values were regenerated. The separate co-factor restraint files (XXX_r_u_LIG.tbl, XXX_l_u_LIG.tbl) are replaced by XXX_hbonds.tbl.
• Obsolete content was removed: structures-matched (superseded by HADDOCK-ready), structures-orig.tgz (the original BM5 files, available from https://zlab.umassmed.edu/benchmark/), symmetries, the HADDOCK 2.4 run and statistics scripts, the original preparation scripts and the old restraint-generation scripts (replaced by haddock-restraints), and the Martini coarse-grain models (XXX_r_u_cg.pdb, XXX_l_u_cg.pdb) with their XXX_cg_to_aa.tbl restraints, which haddock3 generates on the fly (topocg module).