// NewFragment constructs a new fragment from a full query sequence and the
// hit from the HHR file.
//
// Since NewFragment requires access to the raw PDB alpha-carbon atoms (and
// the sequence) of the template hit, you'll also need to pass a path to the
// PDB database. (Which is a directory containing a flat list of all
// PDB files used to construct the corresponding hhblits database.) This
// database is usually located inside the 'pdb' directory contained in the
// corresponding hhsuite database. i.e., $HHLIB/data/pdb-select25/pdb
func NewFragment(
pdbDb PDBDatabase, qs seq.Sequence, hit hhr.Hit) (Fragment, error) {
pdbName := getTemplatePdbName(hit.Name)
pdbEntry, err := pdb.ReadPDB(path.Join(
pdbDb.PDB(), fmt.Sprintf("%s.pdb", pdbName)))
if err != nil {
pdbEntry, err = pdb.ReadPDB(path.Join(
pdbDb.PDB(), fmt.Sprintf("%s.ent.gz", pdbName)))
if err != nil {
return Fragment{}, err
}
}
// Load in the sequence from the PDB file using the SEQRES residues.
ts, te := hit.TemplateStart, hit.TemplateEnd
chain := pdbEntry.Chain(pdbName[4])
if chain == nil {
return Fragment{}, fmt.Errorf("Could not find chain '%c' in PDB "+
"entry '%s'.", pdbName[4], pdbEntry.Path)
}
tseq := seq.Sequence{
Name: pdbName,
Residues: make([]seq.Residue, te-ts+1),
}
// We copy here to avoid pinning pdb.Entry objects.
copy(tseq.Residues, chain.Sequence[ts-1:te])
frag := Fragment{
Query: qs.Slice(hit.QueryStart-1, hit.QueryEnd),
Template: tseq,
Hit: hit,
CaAtoms: nil,
}
// We designate "corrupt" if the query/template hit regions are of
// different length. i.e., we don't allow gaps (yet).
// BUG(burntsushi): Fragments with gaps are marked as corrupt.
if hit.QueryEnd-hit.QueryStart != hit.TemplateEnd-hit.TemplateStart {
return frag, nil
}
// We also designate "corrupt" if there are any gaps in our alpha-carbon
// atom list.
atoms := chain.SequenceCaAtomSlice(ts-1, te)
if atoms == nil {
return frag, nil
}
// One again, we copy to avoid pinning memory.
frag.CaAtoms = make([]structure.Coords, len(atoms))
copy(frag.CaAtoms, atoms)
return frag, nil
}
func createChains(pdbFiles []string) []*pdb.Chain {
chains := make([]*pdb.Chain, 0, len(pdbFiles))
for _, pdbFile := range pdbFiles {
entry, err := pdb.ReadPDB(pdbFile)
util.Warning(err, "Could not open PDB file '%s'", pdbFile)
for _, chain := range entry.Chains {
if !chain.IsProtein() {
continue
}
chains = append(chains, chain)
}
}
return chains
}
func PDBOpen(fpath string) (*pdb.Entry, []*pdb.Chain, error) {
pdbNameParse := func(fpath string) (string, []byte, string) {
dir, base := path.Dir(fpath), path.Base(fpath)
pieces := strings.Split(base, ":")
var idents []byte
base = pieces[0]
if len(pieces) > 2 {
Fatalf("Too many colons in PDB file path '%s'.", fpath)
} else if len(pieces) == 2 {
chains := strings.Split(pieces[1], ",")
idents = make([]byte, len(chains))
for i := range chains {
if len(chains[i]) > 1 {
Fatalf("Chain '%s' is more than one character.", chains[i])
}
idents[i] = byte(chains[i][0])
}
} else if len(base) == 5 { // special case for '{pdb-id}{chain-id}'
idents = []byte{base[4]}
base = base[0:4]
}
if dir == "." {
switch len(base) {
case 4:
return PDBPath(base), idents, base
case 6:
return CathPath(base), idents, base
case 7:
if base[0] == 'd' {
return ScopPath(base), idents, base
} else {
return CathPath(base), idents, base
}
}
}
return path.Join(dir, base), idents, ""
}
fp, idents, idcode := pdbNameParse(fpath)
entry, err := pdb.ReadPDB(fp)
if err != nil {
err = fmt.Errorf("Error reading '%s': %s", fp, err)
return nil, nil, err
}
if len(idcode) > 0 {
if len(idcode) == 6 || (len(idcode) == 7 && idcode[0] != 'd') {
entry.Cath = idcode
} else if len(idcode) == 7 && idcode[0] == 'd' {
entry.Scop = idcode
}
}
var chains []*pdb.Chain
if len(idents) == 0 {
chains = entry.Chains
} else {
chains = make([]*pdb.Chain, 0, 5)
for _, c := range idents {
chain := entry.Chain(c)
if chain == nil {
Warnf("Chain '%c' does not exist for '%s'.",
c, entry.IdCode)
continue
}
chains = append(chains, chain)
}
}
return entry, chains, nil
}
func PDBRead(path string) *pdb.Entry {
entry, err := pdb.ReadPDB(path)
Assert(err, "Could not open PDB file '%s'", path)
return entry
}