def summpbsa():
import numpy as np
import subprocess as sp
from collections import defaultdict
from mdanaly import cmap
ref = "complex_1/em-c1.pdb"
sp.check_output("echo \"1 0 \" | trjconv -f complex_1/em-c1.gro -s complex_1/em-c1.tpr -o complex_1/em-c1.pdb",
shell=True)
CMAP = cmap.ContactMap("./complex_1/em-c1.pdb")
at = CMAP.findAtomType('non-hydrogen', ref)
rd = defaultdict(list)
rd["A"] = range(1, 300)
ld = defaultdict(list)
ld["B"] = range(1, 10)
rndx = CMAP.findAtomNdx(ref, rd, "A", at, False)
lndx = CMAP.findAtomNdx(ref, ld, "B", at, False)
energy = np.loadtxt("./mmpbsa_summary.dat")
f_idx = [int(x) for x in energy[:, 0]]
result = np.array([])
for mmpbsa in energy:
i = int(mmpbsa[0])
sp.check_output(
"echo \"1 0 \" | trjconv -f complex_%d/em-c%d.gro -s complex_%d/em-c%d.tpr -o complex_%d/em-c%d.pdb"
% (i, i, i, i, i, i), shell=True)
fn = ["complex_%d/em-c%d.pdb" % (i, i)]
c = CMAP.cmap_ca(fn, 0.35, False, [rndx, lndx], 0, False,
len(rndx) * len(rndx), perAtom=[False, False],
ccutoff=2.0, NbyN=False)
e = mmpbsa[-1]
if not result.shape[0]:
result = c * (2.71828 ** (-1.0 * e / 2.5))
else:
result += c * (2.71828 ** (-1.0 * e / 2.5))
MAX = np.max(result)
cmap = np.reshape(result, (len(lndx), len(rndx)))
np.savetxt("cmap_energy_normalized.dat", cmap / MAX, delimiter=" ", fmt="%.4f")
def gen_cmap(args):
"""Load a trajectory file and calculate the atom contactmap.
Parameters
----------
args: argparse object
the argument options
Returns
-------
cmap_dat: pd.DataFrame, shape = [ N, M]
the contactmap along the time. N is number of frames,
M is N_res * N_res (N_res is number of residues for cmap
calculations. )
"""
# TODO: add a residue based selection module here
# select atoms indices for distance calculations
atoms_selections = args.select.split()
if len(atoms_selections) == 2:
atoms_selections = atoms_selections
else:
atoms_selections = [atoms_selections[0]] * 2
top = mt.load(args.s).topology
atom_grp_a = top.select("name %s" % atoms_selections[0])
atom_grp_b = top.select("name %s" % atoms_selections[1])
# load the trajectory file by chunks iteratively
print("Iterloading xtc trajectory file ...... ")
trajs = gmxcli.read_xtc(xtc=args.f, top=args.s, chunk=1000, stride=int(args.dt/args.ps))
cmap_dat = np.array([])
print("Computing contactmap ...... ")
for i, traj in enumerate(trajs):
contmap = cmap.ContactMap(traj=traj, group_a=atom_grp_a, group_b=atom_grp_b, cutoff=args.cutoff)
contmap.generate_cmap(shape="array", switch=args.switch)
if i == 0:
cmap_dat = contmap.cmap_
else:
cmap_dat = np.concatenate((cmap_dat, contmap.cmap_), axis=0)
cmap_dat = pd.DataFrame(cmap_dat)
return cmap_dat
def run_coord_number():
"""
Examples:
gmx_coordnum.py -f test_traj_dt1ns.xtc -s reference.pdb -rc " " 1 40
-lc " " 50 70 -o coord_result.csv -atomtype all all -byres True
-v True -cutoff 0.5
"""
startTime = datetime.now()
args = arguments()
resides_a, resides_b = [], []
group_a, group_b = [], []
if args.byres:
ndx = index.PdbIndex(args.s, [args.rc[0]], args.rc[1:])
ndx.resid_mapper()
#print(ndx.resid_mapper_)
s, e = ndx.resid_mt_style(args.rc[0], args.rc[1], args.rc[2])
#print(s, e)
resides_a = np.arange(s, e + 1)
ndx = index.PdbIndex(args.s, [args.lc[0]], args.lc[1:])
ndx.resid_mapper()
#print(ndx.resid_mapper_)
s, e = ndx.resid_mt_style(args.lc[0], args.lc[1], args.lc[2])
resides_b = np.arange(s, e + 1)
verbose(args.v,
"X-axis resids : " + " ".join([str(x) for x in resides_a]))
verbose(args.v,
"Y-axis resids : " + " ".join([str(x) for x in resides_b]))
else:
# TODO: define a way to select atom slices
# define the atom indices for receptor
'''ndx = index.PdbIndex(reference=args.s, atomtype=args.atomtype[0],
resSeq=args.rc[1:],
chain=[args.rc[0]])
ndx.prepare_selection()
ndx.res_index()'''
group_a = index.gen_atom_index(pdbin=args.s,
chain=[args.rc[0]],
atomtype=args.atomtype[0],
resSeq=args.rc[1:],
style="mdtraj")
#print(group_a)
# define the atom indices for ligand
'''ndx = index.PdbIndex(reference=args.s, atomtype=args.atomtype[1],
resSeq=args.lc[1:],
chain=[args.lc[0]])
ndx.prepare_selection()
ndx.res_index()'''
group_b = index.gen_atom_index(pdbin=args.s,
chain=[args.lc[0]],
atomtype=args.atomtype[-1],
resSeq=args.lc[1:],
style="mdtraj")
verbose(args.v, "Atom indexing processing completed ......")
results = np.array([])
verbose(args.v, "Loading trajectory xtc file ...... ")
trajs = gmxcli.read_xtc(args.f,
args.s,
chunk=1000,
stride=int(args.dt / args.ps))
verbose(args.v, "Performing calculations ...... ")
for i, traj in enumerate(trajs):
verbose(args.v,
"Generate coordinate number for chunk #%d trajectory " % i)
if args.byres:
coord = cmap.CmapNbyN(traj,
resids_a=resides_a,
resids_b=resides_b,
cutoff=args.cutoff)
coord.contact_nbyn()
if i == 0:
results = coord.contacts_by_res_
else:
results = np.concatenate((results, coord.contacts_by_res_),
axis=0)
else:
coord = cmap.ContactMap(traj, group_a, group_b, args.cutoff)
coord.coord_num()
if i == 0:
results = coord.coord_number_
else:
results = np.concatenate((results, coord.coord_number_),
axis=0)
results = pd.DataFrame(results)
results.index = np.arange(results.shape[0]) * args.dt
verbose(args.v, "Saving results to output file ...... ")
results.to_csv(args.o,
sep=",",
header=False,
index=True,
float_format="%.1f")
print("Total Time Usage: ")
print(datetime.now() - startTime)
atom_indices = []
for i in [0, 1]:
print("Please select a group for %s: " % sets[i])
for j, gn in enumerate(ndx.groups):
print("%d : %s" % (j, gn))
used_groups.append(ndx.groups[int(input("Your choice: "))])
rec_ndx = [int(x) - 1 for x in ndx.groupContent(used_groups[0])]
lig_ndx = [int(x) - 1 for x in ndx.groupContent(used_groups[1])]
cmaps = np.array([])
# generate cmap now
for traj in trajs:
CMap = cmap.ContactMap(traj, rec_ndx, lig_ndx, cutoff=0.5)
CMap.generate_atom_pairs()
CMap.generate_cmap(switch=True)
c = CMap.cmap_
if cmaps.shape[0] == 0:
cmaps = c
else:
cmaps = np.concatenate((cmaps, c), axis=0)
# run PCA now
pca_obj = pca.PCA()
pca_obj.fit_transform(cmaps[:, 1:])
print(pca_obj.eigvalues_ratio_)
X_transformed = pca_obj.X_transformed_