def RMSD(CompType='traj'):
oshelf = shelve.open(OutShelf)
if CompType == 'traj':
print '\nComputing overall RMSD distribution'
rmsdhist = calc.QuickRMSD()
oshelf['rmsd'] = rmsdhist
if CompType == 'topclust':
print '\nComputing RMSD from top cluster'
pNative = cg.ProteinNCOS(NativePdb)
pClust = cg.ProteinNCOS(oshelf['clustpdb'])
rmsd = pClust.QuickRMSD(pNative)
oshelf['rmsd_topclust'] = rmsd
oshelf.close()
return
def PhiPsiErr(ErrType='traj'):
print 'Calculating Ramachandran plot errors with native struct'
# get native phi, psi
pNative = cg.ProteinNCOS(NativePdb)
Phi_Native, Psi_Native = pNative.GetPhiPsi(RamaType='Generic')
Phi_Native, Psi_Native = cg.TrimDihedrals(Phi_Native,
Psi_Native,
RamaType='Generic',
ResNums=range(pNative.NRes))
Err = np.zeros([pNative.NRes - 1], float)
# errors from top cluster
if ErrType == 'topclust':
ClustPdb = FMT['CLUSTPDB'] % (OutPrefix, TempSet)
if not os.path.isfile(ClustPdb):
raise IOError('Top cluster for %s missing' % Prefix)
p = cg.ProteinNCOS(ClustPdb, Model=1)
PhiErr, PsiErr, Err = pNative.GetPhiPsiDiff(p, RamaType='Generic')
# remove undefined dihedrals and recompute Err
PhiErr = PhiErr[1:]
PsiErr = PsiErr[:-1]
Err = np.sqrt(PhiErr**2. + PsiErr**2.)
# errors from entire traj
if ErrType == 'traj':
calc.RamaChandran()
RamaPickle = FMT['RAMA'] % (OutPrefix, TempSet)
with open(RamaPickle, 'r') as of:
data = pickle.load(of)
Phi, Psi, hist = data['Generic']
for i in range(len(Phi_Native)):
# per frame phi-psi errors
this_Phidiff = sim.geom.NearestAngle(Phi_Native[i] - Phi[:, i],
0.0)
this_Psidiff = sim.geom.NearestAngle(Psi_Native[i] - Psi[:, i],
0.0)
Err[i] = np.mean(
np.sqrt(this_Phidiff**2 +
this_Psidiff**2)) # need to do a np.mean over frames
oshelf = shelve.open(OutShelf)
oshelf['ramaprob'] = RamaPickle
oshelf['ramaerr'] = (Phi_Native, Psi_Native, Err)
oshelf.close()
return
def RamaChandran():
# ramachandran plots from native
print '\nGenerating Ramachandran plots from native'
p = lib.ProteinNCOS(NativePdb)
Phi, Psi = p.GetPhiPsi()
picklename = OutPrefix + '_native.rama.pickle'
with open(picklename, 'w') as of:
pickle.dump((Phi, Psi), of)
# ramachandran plots from entire traj
print '\nRamachandran plots at Temp = %3.2fK' % TempSet
calc.RamaChandran()
def PhiPsiErr(ErrType='traj'):
print '\nCalculating Ramachandran plot errors with native struct'
# get native phi, psi
pNative = cg.ProteinNCOS(NativePdb)
Phi_Native, Psi_Native = pNative.GetPhiPsi()
Err = np.zeros(len(Phi_Native), float)
# errors from top cluster
if ErrType == 'topclust':
ClustPdb = FMT['CLUSTPDB'] % (OutPrefix, TempSet)
if not os.path.isfile(ClustPdb):
raise IOError('Top cluster for %s missing' % Prefix)
p = cg.ProteinNCOS(ClustPdb, Model=1)
PhiErr, PsiErr, Err = pNative.GetPhiPsiDiff(p)
# errors from entire traj
if ErrType == 'traj':
calc.RamaChandran()
RamaPickle = FMT['RAMA'] % (OutPrefix, TempSet)
with open(RamaPickle, 'r') as of:
data = pickle.load(of)
Phi, Psi, hist = data
for i in range(len(Phi_Native)):
# per frame phi-psi errors
this_Phidiff = sim.geom.NearestAngle(Phi_Native[i] - Phi[:, i],
0.0)
this_Psidiff = sim.geom.NearestAngle(Psi_Native[i] - Psi[:, i],
0.0)
Err[i] = np.mean(
np.sqrt(this_Phidiff**2 +
this_Psidiff**2)) # need to do a np.mean over frames
oshelf = shelve.open(OutShelf)
with open(RamaPickle, 'w') as of:
data = pickle.load(RamaPickle)
oshelf['ramaprob'] = data
oshelf['ramaerr'] = (Phi_Native, Psi_Native, Err)
oshelf.close()
return
def makeRamaPlot(TempAA, TempCG):
print 'RAMACHANDRAN PLOTS'
print '------------------'
global c_AA, c_CG
# parse AA and CG filenames
fAA = utils.parseAA(polymer, TempSet = TempAA)
fCG = utils.parseCG(polymer, TempSet = TempCG)
TempAA = fAA['Temp']
TempCG = fCG['Temp']
# get top cluster ramachandran plots
AAClustPdb = FMT['CLUSTPDB'] % (AAPrefix, fAA['Temp'])
if not os.path.isfile(AAClustPdb):
print 'Error: AA top cluster struct missing. Cannot compute top cluster dihderals'
Phi_AA = None
Psi_AA = None
else:
pAA = lib.ProteinNCOS(AAClustPdb)
Phi_AA, Psi_AA = pAA.GetPhiPsi()
Phi_AA, Psi_AA = lib.TrimDihedrals(Phi_AA, Psi_AA, range(pAA.NRes))
CGClustPdb = FMT['CLUSTPDB'] % (CGPrefix, fCG['Temp'])
if not os.path.isfile(CGClustPdb):
print 'Error: CG top cluster struct missing. Cannot compute top cluster dihderals'
Phi_CG = None
Psi_CG = None
else:
pCG = lib.ProteinNCOS(CGClustPdb)
Phi_CG, Psi_CG = pCG.GetPhiPsi()
Phi_CG, Psi_CG = lib.TrimDihedrals(Phi_CG, Psi_CG, range(pCG.NRes))
# get whole traj based ramachandran plots
c_AA.Update(TrajFn = fAA['Traj'], Temp = TempAA)
c_CG.Update(TrajFn = fCG['Traj'], Temp = TempCG)
c_AA.RamaChandran()
c_CG.RamaChandran()
# plot
AApickle = FMT['RAMA'] % (AAPrefix, TempAA)
CGpickle = FMT['RAMA'] % (CGPrefix, TempCG)
with open(AApickle, 'r') as of: phiaa, psiaa, haa = pickle.load(of)['Generic']
with open(CGpickle, 'r') as of: phicg, psicg, hcg = pickle.load(of)['Generic']
pmf_AA = -np.log(haa[1])
pmf_CG = -np.log(hcg[1])
fig, axes = plt.subplots(nrows = 1, ncols = 2, figsize = (10,7), facecolor = 'w', edgecolor = 'w')
ax1 = axes.flat[0]
ax2 = axes.flat[1]
X1 = rad2deg(haa[0][0]) ; Y1 = rad2deg(haa[0][1])
X2 = rad2deg(hcg[0][0]) ; Y2 = rad2deg(hcg[0][1])
im1 = ax1.imshow(np.transpose(pmf_AA), aspect = 'equal', interpolation = 'none', origin = 'lower',
cmap = cm.jet, extent = [X1.min(), X1.max(), Y1.min(), Y1.max()])
im2 = ax2.imshow(np.transpose(pmf_CG), aspect = 'equal', interpolation = 'none', origin = 'lower',
cmap = cm.jet, extent = [X2.min(), X2.max(), Y2.min(), Y2.max()])
#TODO: scatter the top cluster dihedrals on this plot
fig.colorbar(im1, ax=axes.ravel().tolist(), label = 'pmf (kT)', orientation = 'vertical')
ax1.set_title('AA : %3.2f K' % TempAA)
ax2.set_title('CG : %3.2f K' % TempCG)
for ax in [ax1, ax2]:
ax.set_xlabel(r'$\phi$')
ax.set_ylabel(r'$\psi$')
ax.set_xlim([-190, 190])
ax.set_ylim([-190, 190])
ax.axvline(0, ls = '-', lw = 1, color = 'black')
ax.axhline(0, ls = '-', lw = 1, color = 'black')
figname = OutPrefix + '.%3.2f.ramaplot.png' % TempAA
plt.savefig(figname, bbox_inches = 'tight')
#!/usr/bin/env python
import os, numpy as np
import matplotlib
import matplotlib.pyplot as plt
import cgprotein as lib
pdbname = '1l2y'
gly_dir = os.path.abspath('../native_struct/mapped_pseudoGLY')
nogly_dir = os.path.abspath('../native_struct/mapped')
p_gly = lib.ProteinNCOS(os.path.join(gly_dir, pdbname + '.pdb'), hasPseudoGLY = True)
phi_gly, psi_gly = p_gly.GetPhiPsi()
p_nogly = lib.ProteinNCOS(os.path.join(nogly_dir, pdbname + '.pdb'))
phi_nogly, psi_nogly = p_nogly.GetPhiPsi()
fig = plt.figure(figsize = (10, 5), facecolor = 'w', edgecolor = 'w')
ax1 = fig.add_subplot(1,2,1)
ax2 = fig.add_subplot(1,2,2)
ax1.scatter(phi_gly * (180./np.pi), psi_gly * (180./np.pi), c = 'red', s = 50)
ax2.scatter(phi_nogly * (180./np.pi), psi_nogly * (180./np.pi), c = 'red', s = 50)
ax1.set_title('peptide %s, pseudo gly' % pdbname)
ax2.set_title('peptide %s' % pdbname)
for ax in [ax1, ax2]:
ax.set_xlabel(r'$\phi$')
ax.set_ylabel(r'$\psi$')
ax.axhline(0.0, ls = '-', lw = 2, color = 'black')
ax.axvline(0.0, ls = '-', lw = 2, color = 'black')
ax.set_xlim([-180, 180])
ax.set_ylim([-180, 180])
print 'ANALYSING MULTI-PROTEIN FOR %s at %3.2f K' % (Prefix, TempSet) print '----------------------------------------------------------------' # set up Compute object print 'Creating Compute object' calc = cg.Compute(NativePdb = NativePdb, TrajFn = TrajFn, Temp = TempSet, Prefix = OutPrefix, hasPseudoGLY = hasPseudoGLY) # cluster the room temp traj print 'Clustering trajectory' calc.Cluster() # top-cluster contact analysis Trj = pickleTraj(TrajFn) BoxL = Trj.FrameData['BoxL'] ClustPdb = FMT['CLUSTPDB'] % (calc.Prefix, calc.Temp) p = cg.ProteinNCOS(ClustPdb, Model = 1) # top cluster contact map print 'Calculating contact map for top cluster' cmap, cdist = p.GetResContacts(BoxL = BoxL) picklename = FMT['RESCONTACTS'] % (calc.Prefix + '_topclust', calc.Temp) with open(picklename, 'w') as of: pickle.dump(cmap, of) # top cluster dihedrals print 'Calculating dihedral angles for top cluster' Phi, Psi = p.GetPhiPsi(BoxL = BoxL) picklename = FMT['RAMA'] % (calc.Prefix + '_topclust', calc.Temp) with open(picklename, 'w') as of: pickle.dump((Phi, Psi), of) # traj contact map print 'Calculating contact map from entire trajectory'
Traj = {'ala_spc': os.path.abspath('../ala_spc_AA/Lammps/ala_spc.300.00.lammpstrj.gz'),
'ala15': os.path.abspath('../ala15_AA/Lammps/ala15.299.00.lammpstrj.gz'),
'leu15': os.path.abspath('../leu15_AA/Lammps/leu15_wham.407.00.lammpstrj.gz'),
'val15': os.path.abspath('../val15_AA/Lammps/val15_wham.367.00.lammpstrj.gz')
}
Temp = {'ala_spc': 300.0, 'ala15': 299.00, 'leu15': 407.00, 'val15': 367.00}
NativePdb = {'ala_spc': 'ideal_helix.pdb', 'ala15': 'ideal_helix.pdb', 'leu15': 'ideal_helix.pdb', 'val15': hairpin_pdb}
fig = plt.figure(figsize = (22, 15), facecolor = 'w', edgecolor = 'w')
ind = 1
gs = gridspec.GridSpec(3,4)
for i, fftype in enumerate(fftypes):
print '-------- %s --------' % fftype
# get native Ramachandran angles
pNative = cg.ProteinNCOS(NativePdb[fftype])
PhiNative, PsiNative = pNative.GetPhiPsi(RamaType = 'Generic')
PhiNative, PsiNative = cg.TrimDihedrals(PhiNative, PsiNative, ResNums = range(NRes))
PhiNative *= (180 / np.pi)
PsiNative *= (180 / np.pi)
# create compute object
calc = cg.Compute(NativePdb = NativePdb[fftype], TrajFn = Traj[fftype], Temp = Temp[fftype], Prefix = fftype)
# get cluster ramachandran angle probabilities
calc.RamaChandran()
RamaPickle = utils.FMT['RAMA'] % (fftype, Temp[fftype])
with open(RamaPickle, 'r') as of: data = pickle.load(of)['Generic']
(x,y), ramahist, err = data[-1]
x *= (180 / np.pi)
y *= (180 / np.pi)
def Overlay(NativePdb,
Pdb,
OutPrefix=None,
Label='',
SinglePlot=False,
hasPseudoGLY=False):
global doRotate
if OutPrefix is None: OutPrefix = 'go'
# parse BBInds
p_cg = cgprotein.ProteinNCOS(NativePdb, hasPseudoGLY=hasPseudoGLY)
BBInds = p_cg.GetBBInds()
# read pdbs
pNative = protein.ProteinClass(NativePdb)
p = protein.ProteinClass(Pdb, Model=1)
# rotate the native pdb
if doRotate: pNative = RotateProteinClass(pNative)
# align with rotated native struct (produces weird results with vmd, so let vmd do its own )
if not Renderer == 'vmd':
p, pNative = AlignProtein(p, pNative, BBInds)
# write to first set of tmp pdb files
tmpNativePdb = os.path.join(os.getcwd(), '%s_tmpnative.pdb' % OutPrefix)
tmpPdb = os.path.join(os.getcwd(), '%s_tmp.pdb' % OutPrefix)
pNative.WritePdb(tmpNativePdb)
p.WritePdb(tmpPdb)
# now reverse map these pdbs to generate an approximate carbonyl group
# so that STRIDE can assign secondary structures
mapNCOS.ReverseMap(CGPdb=tmpNativePdb,
Prefix=tmpNativePdb.split('.pdb')[0],
hasPseudoGLY=hasPseudoGLY)
mapNCOS.ReverseMap(CGPdb=tmpPdb,
Prefix=tmpPdb.split('.pdb')[0],
hasPseudoGLY=hasPseudoGLY)
# fill dictionary for renderer script
d = {
'TMPNATIVEPDB': tmpNativePdb,
'TMPPDB': tmpPdb,
}
# pymol
if Renderer == 'pymol':
d['FILENAME'] = OutPrefix + '.png' if not SinglePlot else OutPrefix + '_tmp0.png'
tmpPml = OutPrefix + '.pml'
file(tmpPml, 'w').write(s_pymol % d)
cmdstr1 = '%s -Qc %s' % (PYMOLEXEC, tmpPml)
os.system(cmdstr1)
for x in [tmpNativePdb, tmpPdb, tmpPml]:
os.remove(x)
# vmd
elif Renderer == 'vmd':
d['FILEPREFIX'] = OutPrefix if not SinglePlot else OutPrefix + '_tmp0'
d['TACHYONEXEC'] = TACHYONEXEC
tmpTcl = OutPrefix + '.tcl'
file(tmpTcl, 'w').write(s_vmd % d)
cmdstr1 = '%s -dispdev text -eofexit -e %s > /dev/null 2>&1' % (
VMDEXEC, tmpTcl)
os.system(cmdstr1)
for x in [tmpNativePdb, tmpPdb, tmpTcl, d['FILEPREFIX']]:
os.remove(x)
else:
print 'ERROR: Renderer not found'
exit()
# if single plot with supplied labels is requested (mostly when used from the command line)
if SinglePlot:
if not Label:
rmsd = sim.geom.RMSD(pNative.Pos[BBInds], p.Pos[BBInds])
print rmsd
Label = r'$RMSD = %2.2f \AA$' % rmsd
if Renderer == 'pymol': pic0 = OutPrefix + '_tmp0.png'
else: pic0 = OutPrefix + '_tmp0.tga'
pic1 = OutPrefix + '_tmp1.png'
cmdstr2 = '%s %s -trim -bordercolor white -background white -border 50x50 -quality 100 %s' % (
IMAGEMAGICKEXEC, pic0, pic1)
os.system(cmdstr2)
pic = mpimg.imread(pic1)
fig = plt.figure(figsize=(5, 5), facecolor='w', edgecolor='w')
ax = fig.add_subplot(1, 1, 1)
ax.imshow(pic, aspect='auto')
#ax.set_title(Label, fontsize = 8)
ax.set_xticks([])
ax.set_yticks([])
ax.set_xticklabels([])
ax.set_yticklabels([])
figname = OutPrefix + '.png'
plt.savefig(figname, bbox_inches='tight')
for x in [pic0, pic1]:
os.remove(x)
return