void read_pdb(Model *model, const std::string file,
std::vector<std::string>& pdb_file_names,
std::vector<IMP::Particles>& particles_vec,
bool residue_level, bool heavy_atoms_only, int multi_model_pdb,
bool explicit_water) {
IMP::atom::Hierarchies mhds;
IMP::atom::PDBSelector* selector;
if (residue_level) // read CA only
selector = new IMP::atom::CAlphaPDBSelector();
else if (heavy_atoms_only) { // read without hydrogens
if (explicit_water)
selector = new IMP::atom::NonHydrogenPDBSelector();
else
selector = new IMP::atom::NonWaterNonHydrogenPDBSelector();
} else { // read with hydrogens
if (explicit_water)
selector = new IMP::atom::NonAlternativePDBSelector();
else
selector = new IMP::atom::NonWaterPDBSelector();
}
if (multi_model_pdb == 2) {
mhds = read_multimodel_pdb(file, model, selector, true);
} else {
if (multi_model_pdb == 3) {
IMP::atom::Hierarchy mhd =
IMP::atom::read_pdb(file, model, selector, false, true);
mhds.push_back(mhd);
} else {
IMP::atom::Hierarchy mhd =
IMP::atom::read_pdb(file, model, selector, true, true);
mhds.push_back(mhd);
}
}
for (unsigned int h_index = 0; h_index < mhds.size(); h_index++) {
IMP::ParticlesTemp ps =
get_by_type(mhds[h_index], IMP::atom::ATOM_TYPE);
if (ps.size() > 0) { // pdb file
std::string pdb_id = file;
if (mhds.size() > 1) {
pdb_id = trim_extension(file) + "_m" +
std::string(boost::lexical_cast<std::string>(h_index + 1)) +
".pdb";
}
pdb_file_names.push_back(pdb_id);
particles_vec.push_back(IMP::get_as<IMP::Particles>(ps));
if (mhds.size() > 1) {
IMP_LOG_TERSE(ps.size() << " atoms were read from PDB file " << file
<< " MODEL " << h_index + 1 << std::endl);
} else {
IMP_LOG_TERSE(ps.size() << " atoms were read from PDB file " << file
<< std::endl);
}
}
}
}
int main(int argc, char **argv) {
try {
desc.add_options()
("help,h",
"Print messages about any errors encountered in parsing the pdb file.")
("input-file,i", po::value< std::string >(&input),
"input pdb file");
po::positional_options_description p;
p.add("input-file", 1);
po::variables_map vm;
po::store(
po::command_line_parser(argc,
argv).options(desc).positional(p).run(),
vm);
po::notify(vm);
if (vm.count("help") || input.empty()) {
print_help();
return 1;
}
IMP_NEW(IMP::Model, m, ());
m->set_log_level(IMP::SILENT);
IMP::set_log_level(IMP::VERBOSE);
IMP::atom::Hierarchies inhs;
IMP_CATCH_AND_TERMINATE(inhs= IMP::atom::read_multimodel_pdb(input, m));
for (unsigned int i=0; i< inhs.size(); ++i) {
IMP::atom::add_bonds(inhs[i]);
}
return 0;
} catch (const IMP::Exception &e) {
std::cerr << "Error: " << e.what() << std::endl;
return 1;
} catch (const std::exception &e) {
std::cerr << "Error: " << e.what() << std::endl;
}
}
int main(int argc, char **argv) {
try {
std::vector<std::string> argcs = IMP::setup_from_argv(
argc, argv, "Print warnings about pdb file", "input.pdb", -1);
IMP::set_log_level(IMP::WARNING);
for (unsigned int i = 0; i < argcs.size(); ++i) {
IMP_NEW(IMP::Model, m, ());
m->set_log_level(IMP::SILENT);
IMP::atom::Hierarchies inhs;
IMP_CATCH_AND_TERMINATE(inhs =
IMP::atom::read_multimodel_pdb(argcs[i], m));
for (unsigned int i = 0; i < inhs.size(); ++i) {
IMP::atom::add_bonds(inhs[i]);
}
}
return 0;
}
catch (const IMP::Exception &e) {
std::cerr << "Error: " << e.what() << std::endl;
return 1;
}
catch (const std::exception &e) {
std::cerr << "Error: " << e.what() << std::endl;
}
}
void ResidueRotamer::set_coordinates(unsigned index,
IMP::atom::Residue &rd) const {
IMP_USAGE_CHECK(index < size_, "no rotamer at given index");
IMP_USAGE_CHECK(rd.get_residue_type() == residue_type_, "wrong residue type");
IMP::atom::Hierarchies mhs = IMP::atom::get_by_type(rd, IMP::atom::ATOM_TYPE);
for (size_t i = 0; i != mhs.size(); ++i) {
IMP::atom::Atom at = mhs[i].get_as_atom();
IMP::atom::AtomType at_t = at.get_atom_type();
if (get_atom_exists(at_t)) {
const IMP::algebra::Vector3D &coords = get_coordinates(index, at_t);
IMP::core::XYZ(at).set_coordinates(coords);
}
}
}
int main(int argc, char *argv[]) {
IMP::Strings args = IMP::setup_from_argv(argc, argv,
"Score a protein-ligand complex",
"file.mol2 file.pdb [libfile]", -1);
IMP::set_log_level(IMP::SILENT);
std::string mol2name, pdbname;
for (size_t i = 0; i < args.size(); ++i) {
std::string nm(args[i]);
if (nm.rfind(".mol2") == nm.size() - 5) {
mol2name = nm;
} else if (nm.rfind(".pdb") == nm.size() - 4) {
pdbname = nm;
} else {
break;
}
}
if (mol2name.empty() || pdbname.empty()) {
std::cerr << "Usage: " << argv[0] << " file.mol2 file.pdb [libfile]"
<< std::endl;
return EXIT_FAILURE;
}
IMP::TextInput lib;
if (args.size() == 3) {
lib = IMP::TextInput(args[2]);
}
{
int lib_requested = 0;
if (lib) lib_requested++;
if (pose_score) lib_requested++;
if (rank_score) lib_requested++;
if (lib_requested > 1) {
std::cerr << "Can only specify one of --pose, --rank, "
<< "or a library name" << std::endl;
return EXIT_FAILURE;
}
}
IMP_NEW(IMP::Model, m, ());
IMP::atom::Hierarchy p, l;
{
IMP::SetLogState ss(IMP::SILENT);
p = IMP::atom::read_pdb(pdbname, m, new IMP::atom::ATOMPDBSelector());
IMP::atom::add_protein_ligand_score_data(p);
l = IMP::atom::read_mol2(mol2name, m);
IMP::atom::add_protein_ligand_score_data(l);
}
IMP::atom::Hierarchies mols =
IMP::atom::get_by_type(l, IMP::atom::RESIDUE_TYPE);
IMP::Pointer<IMP::atom::ProteinLigandAtomPairScore> ps
= get_pair_score(lib);
double d = ps->get_maximum_distance();
IMP_NEW(IMP::core::GridClosePairsFinder, gcpf, ());
gcpf->set_distance(d);
IMP::ParticlesTemp patoms = IMP::atom::get_leaves(p);
IMP::ParticleIndexes ipatoms = IMP::get_indexes(patoms);
for (unsigned int i = 0; i < mols.size(); ++i) {
// IMP::SetLogState ss(i==0? TERSE: IMP::SILENT);
IMP::ParticlesTemp latoms = IMP::atom::get_leaves(mols[i]);
IMP::ParticleIndexes ilatoms = IMP::get_indexes(latoms);
IMP::ParticleIndexPairs ppt =
gcpf->get_close_pairs(m, ipatoms, ilatoms);
double score = ps->evaluate_indexes(m, ppt, NULL, 0, ppt.size());
std::cout << "Score for " << mols[i]->get_name() << " is " << score
<< std::endl;
}
ps->set_was_used(true);
return EXIT_SUCCESS;
}
void RotamerCalculator::transform(const IMP::atom::Hierarchy &protein,
const IMP::PairScore *score, double thr,
double K, int num_iter) const {
IMP_USAGE_CHECK(protein, "protein must me non-null");
IMP::atom::Hierarchies mhs =
IMP::atom::get_by_type(protein, IMP::atom::RESIDUE_TYPE);
const size_t num_res = mhs.size();
ResidueRotamers rotamers;
rotamers.reserve(num_res);
ResidueRotamer::Boxes3D bb_boxes(num_res);
ResidueRotamer::Boxes3D sc_boxes(num_res);
IMP_LOG_VERBOSE("Computing bounding boxes" << std::endl);
std::vector<ResidueRotamer::Boxes3D> rot_boxes(num_res);
for (size_t i = 0; i != num_res; ++i) {
IMP::atom::Residue rd = mhs[i].get_as_residue();
ResidueRotamer rr = get_rotamer(rd, thr);
rotamers.push_back(rr);
rr.create_bounding_boxes(bb_boxes[i], sc_boxes[i], rot_boxes[i]);
}
IMP_LOG_VERBOSE("Equation (39)" << std::endl);
// implementation of equation (39) E^{BB}_{ij}
std::vector<std::vector<double> > E_bb(num_res);
for (size_t i = 0; i != num_res; ++i) {
IMP_LOG_VERBOSE("Processing residue " << i << " out of " << num_res
<< std::endl);
IMP::atom::Residue rd_i = mhs[i].get_as_residue();
IMP::atom::Hierarchies at_i =
IMP::atom::get_by_type(rd_i, IMP::atom::ATOM_TYPE);
E_bb[i].resize(rotamers[i].get_size());
// for each rotamer j...
for (unsigned j = 1; j < rotamers[i].get_size(); ++j) {
rotamers[i].set_coordinates(j, rd_i);
// for each l in Nbr_BB(i)...
for (size_t l = 0; l != num_res; ++l)
if (l != i) {
if (!ResidueRotamer::intersect(bb_boxes[l], sc_boxes[i])) continue;
IMP::atom::Residue rd_l = mhs[i].get_as_residue();
IMP::atom::Hierarchies at_l =
IMP::atom::get_by_type(rd_l, IMP::atom::ATOM_TYPE);
// for each k in side chain(i)...
for (size_t k = 0; k != at_i.size(); ++k) {
IMP::atom::Atom at_ik = at_i[k].get_as_atom();
if (is_backbone(at_ik.get_atom_type().get_index())) continue;
// for each n in backbone(l)...
for (size_t n = 0; n != at_l.size(); ++n) {
IMP::atom::Atom at_ln = at_l[n].get_as_atom();
if (!is_backbone(at_ln.get_atom_type().get_index())) continue;
E_bb[i][j] += score->evaluate_index(
protein->get_model(), IMP::ParticleIndexPair(
at_ik.get_particle()->get_index(),
at_ln.get_particle()->get_index()),
nullptr);
}
}
}
rotamers[i].set_coordinates(0, rd_i);
}
}
// now we are ready to compute eq. (41) and (42)
std::vector<std::vector<double> > E_P(num_res);
std::vector<std::vector<double> > q(num_res);
for (size_t i = 0; i != num_res; ++i) {
E_P[i].resize(rotamers[i].get_size());
q[i].resize(rotamers[i].get_size());
}
// we will need to iterate num_iter times
for (int iter = 0; iter < num_iter; ++iter) {
IMP_LOG_VERBOSE("Equation (41), iteration " << iter << std::endl);
// equation (41)...
for (size_t i = 0; i != num_res; ++i) {
IMP::atom::Residue rd_i = mhs[i].get_as_residue();
IMP::atom::Hierarchies at_i =
IMP::atom::get_by_type(rd_i, IMP::atom::ATOM_TYPE);
for (unsigned j = 1; j < rotamers[i].get_size(); ++j) {
E_P[i][j] = E_bb[i][j];
for (size_t l = 0; l != num_res; ++l)
if (l != i) {
IMP::atom::Residue rd_l = mhs[l].get_as_residue();
IMP::atom::Hierarchies at_l =
IMP::atom::get_by_type(rd_l, IMP::atom::ATOM_TYPE);
for (unsigned m = 1; m < rotamers[l].get_size(); ++m) {
if (!ResidueRotamer::intersect(rot_boxes[i][j], rot_boxes[l][m]))
continue;
double q_lm = rotamers[l].get_probability(m);
// now compute E^SC_{ijlm} from eq. (40)
double E_SC = 0;
rotamers[i].set_coordinates(j, rd_i);
rotamers[l].set_coordinates(m, rd_l);
// for k in side-chain(i)...
for (size_t k = 0; k != at_i.size(); ++k) {
IMP::atom::Atom at_ik = at_i[k].get_as_atom();
if (is_backbone(at_ik.get_atom_type().get_index())) continue;
// for n in side-chain(l)...
for (size_t n = 0; n != at_l.size(); ++n) {
IMP::atom::Atom at_ln = at_l[n].get_as_atom();
if (is_backbone(at_ln.get_atom_type().get_index())) continue;
E_SC += score->evaluate_index(
protein->get_model(),
IMP::ParticleIndexPair(
at_ik.get_particle()->get_index(),
at_ln.get_particle()->get_index()),
nullptr);
}
}
rotamers[i].set_coordinates(0, rd_i);
rotamers[l].set_coordinates(0, rd_l);
E_P[i][j] += q_lm * E_SC;
}
}
}
}
IMP_LOG_VERBOSE("Equation (42), iteration " << iter << std::endl);
// equation (42)...
for (size_t i = 0; i != num_res; ++i) {
double denom = 0;
for (unsigned k = 1; k < rotamers[i].get_size(); ++k) {
// IMP_LOG_VERBOSE( "E_P[" << i << "][" << k << "] = " << E_P[i][k] <<
// std::endl);
// IMP_LOG_VERBOSE( "P[" << i << "][" << k << "] = " <<
// rotamers[i].get_probability(k) << std::endl);
denom += std::exp(-K * E_P[i][k]) * rotamers[i].get_probability(k);
}
for (unsigned j = 1; j < rotamers[i].get_size(); ++j) {
q[i][j] =
std::exp(-K * E_P[i][j]) * rotamers[i].get_probability(j) / denom;
// IMP_LOG_VERBOSE( "Computed q[" << i << "][" << j << "] = " <<
// q[i][j] << std::endl);
}
}
// store new probabilities
for (size_t i = 0; i != num_res; ++i)
for (unsigned j = 1; j < rotamers[i].get_size(); ++j)
rotamers[i].probabilities_[j] = q[i][j];
}
// now we are ready to transform - choose max probability
IMP_LOG_VERBOSE("Transforming all residues" << std::endl);
for (size_t i = 0; i != num_res; ++i) {
IMP::atom::Residue rd_i = mhs[i].get_as_residue();
// find max probability
unsigned best = 0;
double best_prob = 0;
for (unsigned j = 1; j < rotamers[i].get_size(); ++j)
if (rotamers[i].get_probability(j) > best_prob) {
best_prob = rotamers[i].get_probability(j);
best = j;
}
if (best > 0) {
IMP_LOG_VERBOSE("Setting coordinates for residue " << rd_i << std::endl);
rotamers[i].set_coordinates(best, rd_i);
}
}
}
ResidueRotamer RotamerCalculator::get_rotamer(const IMP::atom::Residue &rd,
double thr) const {
IMP::atom::ResidueType rt = rd.get_residue_type();
ResidueRotamer r(rt);
// the coordinates at index 0 are the original coordinates
IMP::atom::Hierarchies mhs = IMP::atom::get_by_type(rd, IMP::atom::ATOM_TYPE);
for (size_t i = 0; i != mhs.size(); ++i) {
IMP::atom::Atom at = mhs[i].get_as_atom();
r.add_coordinates(at.get_atom_type(), IMP::core::XYZ(at).get_coordinates());
}
r.size_ = 1;
r.probabilities_.push_back(0);
// unknown residue will not be rotated
unsigned r_idx = rt.get_index();
if (r_idx >= residues_.size() || residues_[r_idx].empty()) {
IMP_LOG_VERBOSE("Residue " << rt << " is unknown" << std::endl);
return r;
}
// query the rotamer library for chi-angles (the library stores angles
// in degrees)
const double to_deg = 180 / IMP::PI, from_deg = IMP::PI / 180;
double psi = get_psi_angle(rd) * to_deg, phi = get_phi_angle(rd) * to_deg;
RotamerLibrary::RotamerRange rr = rl_->get_rotamers_fast(rt, phi, psi, thr);
// we need to create fake particles representing rotated atoms
// (for get_dihedral)
IMP_NEW(IMP::Model, model, ());
IMP_NEW(IMP::Particle, p_a, (model));
IMP_NEW(IMP::Particle, p_b, (model));
IMP_NEW(IMP::Particle, p_c, (model));
IMP_NEW(IMP::Particle, p_d, (model));
IMP::core::XYZ xyz_a = IMP::core::XYZ::setup_particle(p_a);
IMP::core::XYZ xyz_b = IMP::core::XYZ::setup_particle(p_b);
IMP::core::XYZ xyz_c = IMP::core::XYZ::setup_particle(p_c);
IMP::core::XYZ xyz_d = IMP::core::XYZ::setup_particle(p_d);
const ResidueData &res_data = residues_[r_idx];
// iterate through all sets of chi angles
for (RotamerLibrary::RotamerIterator p = rr.begin(); p != rr.end(); ++p) {
r.push_coordinates();
const RotamerAngleTuple &ra = *p;
r.probabilities_.push_back(ra.get_probability());
double desired_angles[4] = {
ra.get_chi1() * from_deg, ra.get_chi2() * from_deg,
ra.get_chi3() * from_deg, ra.get_chi4() * from_deg};
// iterate through chi-angles
for (int i = 0; i < res_data.n_angles; ++i) {
// find the axis atoms
// sometimes the required atoms are missing - log and skip the rotation
// in these cases
bool complete = true;
IMP::atom::Atom at_a = IMP::atom::get_atom(rd, res_data.at_axes[i]);
if (!at_a) {
IMP_LOG_VERBOSE("Residue " << rt << ": no atom of type "
<< res_data.at_axes[i] << std::endl);
complete = false;
}
IMP::atom::Atom at_b = IMP::atom::get_atom(rd, res_data.at_axes[i + 1]);
if (!at_b) {
IMP_LOG_VERBOSE("Residue " << rt << ": no atom of type "
<< res_data.at_axes[i + 1] << std::endl);
complete = false;
}
IMP::atom::Atom at_c = IMP::atom::get_atom(rd, res_data.at_axes[i + 2]);
if (!at_c) {
IMP_LOG_VERBOSE("Residue " << rt << ": no atom of type "
<< res_data.at_axes[i + 2] << std::endl);
complete = false;
}
IMP::atom::Atom at_d = IMP::atom::get_atom(rd, res_data.at_axes[i + 3]);
if (!at_d) {
IMP_LOG_VERBOSE("Residue " << rt << ": no atom of type "
<< res_data.at_axes[i + 3] << std::endl);
complete = false;
}
if (!complete) break;
// get the current coordinates of the axis atoms
IMP::algebra::Vector3D &c_a = r.get_coordinates(at_a.get_atom_type());
IMP::algebra::Vector3D &c_b = r.get_coordinates(at_b.get_atom_type());
IMP::algebra::Vector3D &c_c = r.get_coordinates(at_c.get_atom_type());
IMP::algebra::Vector3D &c_d = r.get_coordinates(at_d.get_atom_type());
xyz_a.set_coordinates(c_a);
xyz_b.set_coordinates(c_b);
xyz_c.set_coordinates(c_c);
xyz_d.set_coordinates(c_d);
// compute the original dihedral angle
double actual_angle = IMP::core::get_dihedral(xyz_a, xyz_b, xyz_c, xyz_d);
// now perform the rotations (about c_b->c_c axis) and store the
// coordinates
int mask = 1 << i;
IMP::algebra::Vector3D rot_axis = c_c - c_b;
IMP::algebra::Rotation3D rot = IMP::algebra::get_rotation_about_axis(
rot_axis, desired_angles[i] - actual_angle);
IMP::algebra::Transformation3D t =
IMP::algebra::get_rotation_about_point(c_b, rot);
for (size_t k = 0; k != mhs.size(); ++k) {
IMP::atom::Atom at = mhs[k].get_as_atom();
IMP::atom::AtomType at_type = at.get_atom_type();
unsigned at_idx = at_type.get_index();
if (at_idx < res_data.rot_atoms.size() &&
(res_data.rot_atoms[at_idx] & mask)) {
// Atom "at" participates in this rotation
IMP::algebra::Vector3D ¤t = r.get_coordinates(at_type);
current = t.get_transformed(current);
}
}
}
}
return r;
}