func init() {
flag.StringVar(&flagChain, "chain", flagChain,
"This may be set to one or more chain identifiers. Only amino acids "+
"belonging to a chain specified will be included.")
flag.StringVar(&flagSplit, "split", flagSplit,
"When set, each FASTA entry produced will be written to a file in the "+
"specified directory with the PDB id code and chain identifier as "+
"the name.")
util.FlagParse("in-pdb-file [out-fasta-file]", "")
if util.NArg() != 1 && util.NArg() != 2 {
util.Usage()
}
}
func main() {
lib = util.StructureLibrary(util.Arg(0))
pdbEntry := util.PDBRead(util.Arg(1))
if util.NArg() == 2 {
for _, chain := range pdbEntry.Chains {
atoms := chain.CaAtoms()
bestFragsForRegion(chain, atoms, 0, len(atoms))
}
} else {
chainId := util.Arg(2)
chain := pdbEntry.Chain(chainId[0])
if chain == nil || !chain.IsProtein() {
util.Fatalf("Could not find protein chain with id '%c'.", chainId)
}
atoms := chain.CaAtoms()
if util.NArg() == 3 {
bestFragsForRegion(chain, atoms, 0, len(atoms))
} else {
if util.NArg() != 5 {
log.Println("Both a start and end must be provided.")
util.Usage()
}
s, e := util.Arg(3), util.Arg(4)
sn, en := util.ParseInt(s)-1, util.ParseInt(e)
if en-sn < lib.FragmentSize() {
util.Fatalf("The range [%s, %s] specifies %d alpha-carbon "+
"atoms while at least %d alpha-carbon atoms are required "+
"for the given fragment library.",
s, e, en-sn, lib.FragmentSize())
}
bestFragsForRegion(chain, atoms, sn, en)
}
}
}
func init() {
flag.Float64Var(&flagThreshold, "threshold", flagThreshold,
"Set the distance threshold to use when computing AUC.")
util.FlagUse("cpu", "cpuprof")
util.FlagParse(
"cath-domain-labels best-of-all-matrix"+
"(bowdb | matrix-file) out-file "+
"[ (bowdb | matrix-file) out-file ... ]",
"Computes the AUC of each aligner matrix (or BOW database) given\n"+
"with respect to the 'best-of-all' matrix given. Each AUC is\n"+
"written to a separate out-file. The sizes of all matrices must\n"+
"be exactly equivalent.\n"+
"Files are interpreted as BOW databases if they have a '.bowdb'\n"+
"file extension.")
util.AssertLeastNArg(4)
if util.NArg()%2 != 0 {
util.Fatalf("There must be an out file for each matrix or bowdb file.")
}
}
func main() {
if len(util.FlagCpuProf) > 0 {
f := util.CreateFile(util.FlagCpuProf)
pprof.StartCPUProfile(f)
defer f.Close()
defer pprof.StopCPUProfile()
}
// Read all CATH domains, the best-of-all matrix, and the matrix for
// each aligner.
domains := readDomains(util.Arg(0))
boa := readMatrix(domains, util.Arg(1))
aligners := make([]aligner, 0)
flibs := make([]flib, 0)
for i := 2; i < util.NArg(); i += 2 {
fpath := util.Arg(i)
if path.Ext(fpath) == ".bowdb" {
db := util.OpenBowDB(fpath)
records, err := db.ReadAll()
util.Assert(err)
bowed := make([]bow.Bowed, domains.in.Len())
for _, b := range records {
if !domains.in.Exists(b.Id) {
util.Fatalf("Found ID in bowdb that isn't in the list "+
"of CATH domains provided: %s", b.Id)
}
bowed[domains.in.Atom(b.Id)] = b
}
flibs = append(flibs, flib{db, bowed, util.Arg(i + 1)})
} else {
aligners = append(aligners, aligner{
readMatrix(domains, fpath),
util.Arg(i + 1),
})
}
}
// Now remove CATH domains that don't have a corresponding structure file.
// We don't do this initially since the matrix files are indexed with
// respect to all CATH domains (includings ones without structure).
// This is an artifact of the fact that the matrices were generated with
// a very old version of CATH.
domains.removeOldDomains()
if a := matrixAuc(domains, boa, boa, flagThreshold); a != 1.0 {
util.Fatalf("Something is wrong. The AUC of the best-of-all matrix "+
"with respect to itself is %f, but it should be 1.0.", a)
}
if len(aligners) > 0 {
fmt.Println("Computing AUC for aligners...")
writeAuc := func(aligner aligner) struct{} {
w := util.CreateFile(aligner.outpath)
a := matrixAuc(domains, boa, aligner.matrix, flagThreshold)
fmt.Fprintf(w, "%f\n", a)
return struct{}{}
}
fun.ParMap(writeAuc, aligners)
}
if len(flibs) > 0 {
fmt.Println("Computing AUC for bowdbs...")
writeAuc := func(flib flib) struct{} {
w := util.CreateFile(flib.outpath)
a := flibAuc(domains, boa, flib, flagThreshold)
fmt.Fprintf(w, "%f\n", a)
return struct{}{}
}
fun.ParMap(writeAuc, flibs)
}
}
func main() {
libFile := util.Arg(0)
brkFile := util.Arg(1)
lib := util.StructureLibrary(libFile)
stderrf("Loading PDB files into memory...\n")
entries := make([]*pdb.Entry, util.NArg()-2)
for i, pdbfile := range flag.Args()[2:] {
entries[i] = util.PDBRead(pdbfile)
}
stderrf("Comparing the results of old fragbag and new fragbag on " +
"each PDB file...\n")
for _, entry := range entries {
stderrf("Testing %s...\n", entry.Path)
fmt.Printf("Testing %s\n", entry.Path)
// Try to run old fragbag first. The output is an old-style BOW.
oldBowStr, err := runOldFragbag(brkFile, entry.Path, lib.Size(),
lib.FragmentSize())
if err != nil {
fmt.Println(err)
fmt.Printf("The output was:\n%s\n", oldBowStr)
divider()
continue
}
oldBow, err := bow.NewOldStyleBow(lib.Size(), oldBowStr)
if err != nil {
fmt.Printf("Could not parse the following as an old style "+
"BOW:\n%s\n", oldBowStr)
fmt.Printf("%s\n", err)
divider()
continue
}
// Now use package fragbag to compute a BOW.
var newBow bow.Bow
if flagOldStyle {
newBow = oldStyle{entry}.StructureBow(lib)
} else {
newBow = newStyle{entry}.StructureBow(lib)
}
// Create a diff and check if they are the same. If so, we passed.
// Otherwise, print an error report.
diff := bow.NewBowDiff(oldBow, newBow)
if diff.IsSame() {
fmt.Println("PASSED.")
divider()
continue
}
// Ruh roh...
fmt.Println("FAILED.")
fmt.Printf("\nOld BOW:\n%s\n\nNew BOW:\n%s\n", oldBow, newBow)
fmt.Printf("\nDiff:\n%s\n", diff)
divider()
}
stderrf("Done!\n")
}