def selectProteinAtomsCrds(self, zrange=[0, 1.0]):
"""get protein atoms' coordinates if the protein
atoms z coordinates in zrange
Parameters
----------
zrange : iterable, length = 2
up and low boundaries
Returns
-------
selected_atoms : np.ndarray, shape = [ N, 3]
the xyz coordinates of selected atoms
N is number of atoms selected
"""
pdbio = parsePDB()
atominfor = pdbio.atomInformation(self.pdb)
ndx = atominfor.keys()
# a list of atom index, strings
selected_ndx = [x for x in ndx if atominfor[x][1] == "Protein"]
cpdb = coordinatesPDB()
crds = np.asarray(cpdb.getAtomCrdByNdx(self.pdb, selected_ndx))
selected_crds = [
x for x in crds if ((x[2] > zrange[0]) and (x[2] < zrange[1]))
]
return np.asarray(selected_crds)
def parsePDB(self):
pio = pdbIO.parsePDB(inPDB=self.inpdb)
infor = pio.atomInformation(self.inpdb)
atomndx = infor.keys()
elements = [infor[x][7] for x in atomndx]
return (atomndx, elements)
def domainCOM(self,
domainf,
ref,
pdbchain='A',
output="com_domains.day",
atomNames=["CA"]):
"""
calculate centor of mass, or geometry center of a domain
:param domainf:str, domain information data file
:param ref: str, reference pdb file
:param pdbchain: chain id
:param output: str, output file name
:return: list of list, com of domains, dimension N*3
"""
dom = parsePDB()
domains = dom.readDomainRes(domainf)
dnames = [x[0] for x in domains]
pdbc = coordinatesPDB()
coms = []
for i in range(len(domains)):
# get atom index of residues in a domain
atomindex = index.gen_atom_index(ref, [
pdbchain,
], domains[i][1:], atomNames, "original")
atomindex = [str(x) for x in atomindex]
# get crds of a list of atoms
crds = pdbc.getAtomCrdByNdx(ref, atomindex)
# calculate geometry center, not centor of mass
com = np.mean(np.asarray(crds), axis=0)
coms.append(list(com))
np.savetxt(output,
np.asarray(coms),
fmt="%8.3f",
delimiter=" ",
header=dnames,
comments="#")
return coms
def elementCount(self, ligand):
from dockml import pdbIO
elements = self.getVdWParams().keys()
atominfor = pdbIO.parsePDB(ligand).atomInformation(ligand)
elem_count = dict(zip(elements, np.zeros(len(elements))))
for atom in atominfor.keys():
if atominfor[atom][7] not in elements:
elem_count["DU"] += 1
else:
elem_count[atominfor[atom][7]] += 1
return elem_count
def resInDomains(self, domainf, residues):
'''
input a list of residues and domain information file
output the ratio of residues in each domain
:param domainf:
:param residues: list, a list of residues from community analysis
:return:
'''
pdb = pdbIO.parsePDB()
dinfor = pdb.readDomainRes(domainf)
# eg. { domain_name: [1, 3, 5]}
domains = collections.defaultdict(list)
d_count = {}
for d in dinfor :
domains[d[0]] = range(d[1], d[2]+1)
d_count[d[0]] = 0
for d in d_count.keys() :
# calculate how many res (in parameter residues) in a domain
d_count[d] = len(set(domains[d]).intersection(set(residues)))
# ratio_outof means how much res in the list (residues) in different domains,
# sum them up, you should get 1
# eg, this list of residues is 100 residues, only 25 in domain HNH,
# therefore, the ratio for HNH is 25%.
ratio_outof = {}
# ratio_indomain means, for a specific domain,
# some ratio of all res in this specific domain, is in the list of residues
# eg. Domain HNH (have 250 residues), in this community residues list, 25 res in
# domain HNH, thus the ratio for HNH is 10%
ratio_indomain = {}
for d in d_count.keys() :
ratio_outof[d] = d_count[d] / float(len(residues))
ratio_indomain[d] = d_count[d] / float(len(domains[d]))
ratio_outof = sorted(ratio_outof.items(), key=lambda x: x[1], reverse=True)
ratio_indomain = sorted(ratio_indomain.items(), key=lambda x:x[1], reverse=True)
return ratio_indomain, ratio_outof
def domainWiseEigVec(self,
domainf,
vectors,
scalefactor=1.0,
output='aver-vectors.dat'):
'''
averaging the vectors on each atoms in a domain
:param domainf: str, domain information data file
:param vectors: list of lists, dimension N*3
:param output: str, output file name
:return: tuple, (domain_average_vectors, domain names)
'''
dom = parsePDB()
domains = dom.readDomainRes(domainf)
# domain_names
d_name = [x[0] for x in domains]
minResIndx = min(sum([x[1:] for x in domains], []))
aver_vec = []
for i in range(len(domains)):
#print(domains[i])
# becasue vectors index starting from 0
resindexlist = []
for k in range((len(domains[i]) - 1) / 2):
#print(domains[i][k*2+1], domains[i][k*2+2])
resindexlist += list(
np.asarray(
range(domains[i][k * 2 + 1], domains[i][k * 2 + 2])) -
minResIndx)
v = self.averageVectors(vectors, resindexlist) * scalefactor
aver_vec.append(v)
np.savetxt(output,
np.asarray(aver_vec),
delimiter=' ',
fmt='%12.5f',
header=" ".join(d_name),
comments="#")
return (aver_vec, d_name)
def __init__(self, ligandPDB, receptorPDB, gridsize=1.0):
'''
:param gridsize: float, grid size, unit angstrom
'''
self.gridsize = gridsize
self.ligpdb = ligandPDB
self.recpdb = receptorPDB
bf = BindingFeature()
self.vdwparm = bf.getVdWParams()
self.eleparm = bf.getElementParams()
self.atominfor = pio.parsePDB().atomInformation(self.ligpdb)
try:
with open(self.ligpdb) as lines:
self.ligndx = [x.split()[1] for x in lines if "ATOM" in x]
except:
self.ligndx = []
def main():
mtxh = MatrixHandle()
args, unknown = arguments()
if args.opt in ["merge", "pair-t-test", "ind-t-test"]:
if args.ds in ['xyz', 'XYZ', '3d']:
data1 = mtxh.loadxyz(args.dat[0],
args.dtype,
args.xyzcol,
xyshift=args.xyshift)
data2 = mtxh.loadxyz(args.dat[1],
args.dtype,
args.xyzcol,
xyshift=args.xyshift)
elif args.ds in ['matrix', 'mtx']:
data1 = mtxh.reshapeMtx(args.dat[0],
args.dtype,
xyshift=args.xyshift)
data2 = mtxh.reshapeMtx(args.dat[1],
args.dtype,
xyshift=args.xyshift)
else:
print("Error: Data-shape is not specified. ")
data1, data2 = np.array([]), np.array([])
if args.opt == "merge":
merged = mtxh.merge_matrix(data1, data2)
np.savetxt(args.out, merged, fmt="%.3f", delimiter=" ")
print("Merge matrix file completed!")
elif args.opt == "pair-t-test":
t, p = stats.ttest_rel(data1[:, 2], data2[:, 2])
print("T statistics: %6.3f " % t)
print("P value : %8.6f" % p)
elif args.opt == "ind-t-test":
t, p = stats.ttest_ind(data1[:, 2], data2[:, 2])
print("T statistics: %6.3f " % t)
print("P value : %8.6f" % p)
fit = np.polyfit(data1[:, 2], data2[:, 2], deg=1)
x, y = data1[:, 2], data2[:, 2]
plt.plot(x, fit[0] * x + fit[1], color='red', lw=2.5)
plt.scatter(x, y, c='b')
plt.xlabel("X")
plt.ylabel("Y")
plt.show()
elif args.opt == "transform":
if args.ds in ['xyz', 'XYZ', '3d']:
data = mtxh.loadxyz(args.dat[0],
args.dtype,
args.xyzcol,
xyshift=args.xyshift)
odata = mtxh.xyz2matrix(data)
np.savetxt(args.out, odata, fmt="%.5f", delimiter=" ")
else:
data = mtxh.reshapeMtx(args.dat[0],
args.dtype,
xyshift=args.xyshift)
np.savetxt(args.out, data, fmt="%.5f", delimiter=" ")
print("Transform matrix file type completed!")
elif args.opt == "extract":
if args.ds in ['xyz', 'XYZ', '3d']:
data = mtxh.loadxyz(args.dat[0],
args.dtype,
args.xyzcol,
xyshift=args.xyshift)
elif args.ds in ['matrix', 'mtx']:
data = mtxh.reshapeMtx(args.dat[0],
args.dtype,
xyshift=args.xyshift)
else:
sys.exit(0)
data = data.astype(np.float)
d = mtxh.extractDomainData(data, args.xyrange[:2], args.xyrange[2:])
np.savetxt(args.out, d, fmt="%.5f")
print("Extract matrix file completed!")
elif args.opt == "average":
aver_data = np.array([])
for i in range(len(args.dat)):
if args.ds in ['xyz', 'XYZ', '3d']:
data = mtxh.loadxyz(args.dat[0],
args.dtype,
args.xyzcol,
xyshift=args.xyshift)
elif args.ds in ['matrix', 'mtx']:
data = mtxh.reshapeMtx(args.dat[0],
args.dtype,
xyshift=args.xyshift)
else:
sys.exit(0)
print(data.shape)
if i == 0:
aver_data = data
else:
aver_data[:, 2] += data[:, 2]
aver_data[:, 2] = aver_data[:, 2] / float(len(args.dat))
np.savetxt(args.out, aver_data, fmt="%.5f")
print("Average matrix files completed!")
elif args.opt == "domain-aver":
if args.ds in ['xyz', 'XYZ', '3d']:
data = mtxh.loadxyz(args.dat[0],
args.dtype,
args.xyzcol,
xyshift=args.xyshift)
elif args.ds in ['matrix', 'mtx']:
data = mtxh.reshapeMtx(args.dat[0],
args.dtype,
xyshift=args.xyshift)
else:
sys.exit(0)
data = data.astype(np.float)
data[:, 0] = data[:, 0] + args.start_res
data[:, 1] = data[:, 1] + args.start_res
drange = []
if os.path.exists(args.domain):
pdb = pdbIO.parsePDB()
domains = pdb.readDomainRes(args.domain)
drange = [x[1:] for x in domains]
else:
drange = [float(x) for x in args.drange]
tofile = open(args.out, 'w')
for i in range(len(drange)):
tofile.write("# %d %s \n" %
(i, " ".join([str(x) for x in drange[i]])))
print(drange)
for i in range(len(drange)):
for j in range(len(drange)):
if args.dzero and i == j:
tofile.write("%3d %3d 0.0 \n" % (i, j))
else:
d = []
#print(len(drange[i])/2)
for r1 in range(int(len(drange[i]) / 2)):
for r2 in range(int(len(drange[j]) / 2)):
ccc = mtxh.extractDomainData(
data,
xrange=drange[i][2 * r1:2 * r1 + 2],
yrange=drange[j][2 * r2:2 * r2 + 2])[:, 2]
d += list(ccc)
tofile.write("%3d %3d %12.3f \n" % (i, j, np.mean(d)))
tofile.close()
print("Domain-wise matrix averaging completed!")
elif args.opt == 'neib0':
if args.ds in ['xyz', 'XYZ', '3d']:
data = mtxh.loadxyz(args.dat[0],
args.dtype,
args.xyzcol,
xyshift=args.xyshift)
elif args.ds in ['matrix', 'mtx']:
data = mtxh.reshapeMtx(args.dat[0],
args.dtype,
xyshift=args.xyshift)
else:
sys.exit(0)
data = data.astype(np.float)
print(data.shape)
newd = mtxh.neiborhood2zero(data,
neiborsize=args.neibsize,
outtype='mtx',
zscale=args.zscale)
np.savetxt(args.out, newd, fmt='%.3f')
def cmap_nbyn(trajs,
ref,
rc,
lc,
v=True,
cutoff=0.35,
allchains=" ABCDEFGH",
atomtype=["sidechain", "sidechain"]):
"""Generate sidechain based contact maps.
Parameters
----------
trajs : list of mt.trajectory objects, shape = N
The input chunks of trajectories, N is number of chunks
ref : str
The reference pdb file name
rc : list, shape = 3
The residue and chain identifier for x-axis
lc : list, shape = 3
The residue and chain identifier for y-axis
cutoff : float, default = 0.35
The distance cutoff, in unit nanometer
v : bool, default = True
Whether print detail information during the calculation
allchains : str, default = 'ABCDEFGH'
All available chain identifiers in the reference pdb files
Returns
-------
"""
pdb = pdbIO.parsePDB(inPDB=ref)
all_resids = pdb.getNdxForRes(ref, chains=allchains)
print(all_resids)
# for chain_a
resids_a = []
for i, item in enumerate(all_resids):
if item[2] in rc[0] and \
int(item[1]) in np.arange(int(rc[1]), int(rc[2])+1):
resids_a.append(i)
print(resids_a)
# for chain_b
resids_b = []
for i, item in enumerate(all_resids):
if item[2] in lc[0] and \
int(item[1]) in np.arange(int(lc[1]), int(lc[2])+1):
resids_b.append(i)
print(resids_b)
contact_map = np.array([])
for i, traj in enumerate(trajs):
# calculate cmap information
verbose(v, "Generate cmap for chunk %5d ......" % i)
contmap = CmapNbyN(traj,
resids_a=resids_a,
resids_b=resids_b,
cutoff=cutoff)
contmap.cmap_nbyn(atomtype=atomtype)
if i == 0:
contact_map = contmap.cmap_
else:
contact_map = np.concatenate((contact_map, contmap.cmap_), axis=0)
return contact_map
def drawTimeSeries2D(
self,
cmapmatrix,
refpdb=[],
fsize=14,
xlabel="",
ylabel="",
cmaptype="Grey",
xlim=[],
ylim=[],
yticks_loc=[],
yticks_labels=[],
yticks_showchainid=False,
xticks_loc=[],
xticks_labels=[],
colorbar_label="",
colorbar_show=False,
savefig="",
):
"""
plot the ligand protein interactions (time series, x axis)
:param cmapmatrix: str, the data file, containing time series matrix file
:param refpdb: list, [ pdbfilename, chain id, residue sequence shift-by ]
:param fsize:
:param xlabel:
:param ylabel:
:param cmaptype:
:param xlim:
:param ylim:
:param yticks_loc:
:param yticks_labels:
:param yticks_showchainid: whether show chainid of protein residues
:param xticks_loc:
:param xticks_lables:
:return:
"""
# load cmap file
cmapdata = np.loadtxt(cmapmatrix, delimiter=",")
cm_sorted = sorted(list(cmapdata), key=lambda x: x[0], reverse=False)
# get key protein residues involving protein ligand binding
key_res = []
true_res = np.sum(np.asarray(cm_sorted)[:, 1:], axis=0) > 0
for res in range(true_res.shape[0]):
if true_res[res]:
key_res.append(res)
print("KEY RES ", key_res)
# get full residue name index list
res_labels = []
if len(refpdb) == 3:
ppdb = pdbIO.parsePDB("")
fullreslist = ppdb.getNdxForRes(refpdb[0], [refpdb[1]])
shortresmap = ppdb.longRes2ShortRes()
fullreslist = [x for x in fullreslist if x[2] in refpdb[1]]
for resk in key_res:
if fullreslist[resk][0] in shortresmap.keys():
resseq = str(resk + refpdb[2])
resname = shortresmap[fullreslist[resk][0]]
chainid = fullreslist[resk][2]
if yticks_showchainid:
id = resname + resseq + chainid
else:
id = resname + resseq
res_labels.append(id)
# only keep the important residue cmap
keyres_cmap = np.asarray(cm_sorted)[:, 1:][:, list(key_res)]
# get the length of x and y axis
shapex = len(key_res)
shapey = cmapdata.shape[0] + 1
print("Protein residue numbers: %d" % shapex)
print("Time point numbers: %d" % shapey)
z = np.transpose(keyres_cmap[:, 1:]).T
plt.pcolormesh(z.T, cmap=plt.get_cmap(cmaptype))
if colorbar_show:
plt.colorbar(label=colorbar_label)
plt.xlabel(xlabel, fontsize=fsize)
plt.ylabel(ylabel, fontsize=fsize)
if len(yticks_loc) and len(yticks_labels):
plt.yticks(yticks_loc, yticks_labels)
else:
if len(refpdb) == 3:
plt.yticks(np.array(range(shapex)) + 0.5, res_labels)
if len(xlim):
plt.xlim(xlim)
if len(ylim):
plt.ylim(ylim)
else:
plt.ylim([0, shapex + 0.5])
if len(xticks_labels) and len(xticks_loc):
plt.xticks(xticks_loc, xticks_labels)
if len(savefig):
plt.savefig(savefig, dpi=2000)
plt.show()
return 1
