def _make_groups(sites, nmol, mol):
# Load the protein template to obtain residue information
template = gpcr_lib.load_template(mol)
residues = np.array(template.residues)
codes = np.array(template.codes)
names = np.array(template.names)
# Setup sites/Groups
maxmem = 0
grouplbl = []
for site in sites:
sidx = [template.residues.index(int(r)) for r in site.split(",")]
maxmem = max(maxmem,len(sidx))
grouplbl.append(",".join(["%s%d"%(codes[idx],residues[idx]) for idx in sidx]))
groups = np.zeros((maxmem,len(sites)*nmol),dtype=int)-1
offset = 0
groupi = 0
for site in sites:
ntake = len(site.split(","))
for mi in range(nmol):
gidx = [offset+mi+nmol*i for i in range(ntake)]
groups[:len(gidx),groupi] = np.asarray(gidx)
groupi += 1
offset += ntake*nmol
return groups, grouplbl
def _make_helical(state,mol) :
"""
Convert a residual state matrix to a
helical state matrix by looking at the residues
in the helices.
Parameters
----------
state : nxR Numpy array
the state matrix,
n is the number of snapshots
R is the number of residues
mol : string
the identifier of the molecule
Returns
-------
nxH Numpy array
the helical state matrix, H is the number of helices
"""
helices = gpcr_lib.load_template(mol).rhelices
hstate = np.zeros([state.shape[0],len(helices)],dtype=np.uint8)
for i,h in enumerate(helices) :
hstate[:,i] = np.any(state[:,h[0]-1:h[1]],axis=1)
return hstate
def _make_groups(sites, nmol, mol):
# Load the protein template to obtain residue information
template = gpcr_lib.load_template(mol)
residues = np.array(template.residues)
codes = np.array(template.codes)
names = np.array(template.names)
# Setup sites/Groups
maxmem = 0
grouplbl = []
for site in sites:
sidx = [template.residues.index(int(r)) for r in site.split(",")]
maxmem = max(maxmem, len(sidx))
grouplbl.append(",".join(
["%s%d" % (codes[idx], residues[idx]) for idx in sidx]))
groups = np.zeros((maxmem, len(sites) * nmol), dtype=int) - 1
offset = 0
groupi = 0
for site in sites:
ntake = len(site.split(","))
for mi in range(nmol):
gidx = [offset + mi + nmol * i for i in range(ntake)]
groups[:len(gidx), groupi] = np.asarray(gidx)
groupi += 1
offset += ntake * nmol
return groups, grouplbl
def _resjointcontacts(filename,label,mat,repeats,out,mol) :
"""
Main analysis routine
Parameters
----------
filename : string
file to analyse
label : string
label for the group
mat : string
matrix identifier
repeats : list of string
replacement pattern for multiple repeats
out : string
output prefix
mol : string
protein identifier
"""
# Load the protein template to obtain residue information
template = gpcr_lib.load_template(mol)
residues = np.array(template.residues)
residues0 = np.arange(1,residues.shape[0]+1)
codes = np.array(template.codes)
names = np.array(template.names)
pcontacts = []
npz = np.load(filename)
pjoint0 = npz["joint"+mat]
pjoint = np.zeros([len(repeats),pjoint0.shape[0],pjoint0.shape[0]])
pjoint[0,:,:] = pjoint0
# Do the same for multiple repeats and average over them
if repeats is not None :
for ri,r in enumerate(repeats[1:],1) :
filename2 = filename.replace(repeats[0],r)
npz = np.load(filename2)
pjoint[ri,:,:] = npz["joint"+mat]
pjoint_std = pjoint.std(axis=0)/np.sqrt(len(repeats))
f2d = plt.figure(2,tight_layout=True)
C = gpcr_lib.draw_joint2d(f2d.gca(),residues0,residues,pjoint_std)
f2d.colorbar(C)
f2d.savefig("%s_%s_2d_std.png"%(out,label),format="png")
# Draw a 2D residue-residue joint probability plot
f2d = plt.figure(1,tight_layout=True)
C = gpcr_lib.draw_joint2d(f2d.gca(),residues0,residues,pjoint.mean(axis=0))
f2d.colorbar(C)
f2d.gca().text(1.04,1.01,"p(A,B)",transform=f2d.gca().transAxes)
f2d.savefig("%s_%s_2d.png"%(out,label),format="png")
f2d = plt.figure(3,tight_layout=True)
C = gpcr_lib.draw_joint2d(f2d.gca(),residues0,residues,pjoint.mean(axis=0),logit=True)
f2d.colorbar(C)
f2d.gca().text(1.02,1.02,"ln p(A,B)",transform=f2d.gca().transAxes)
f2d.savefig("%s_%s_2d_log.png"%(out,label),format="png")
def _rescontacts(filenames, labels, repeats, out, mol, time, every):
"""
Main analysis routine
Parameters
----------
filenames : list of strings
the group of files to analyse
labels : list of string
the label for each group
repeats : list of string
replacement pattern for multiple repeats within each group
out : string
output prefix
mol : string
protein identifier
time : float
the total simulation time
every : int
the reading frequency
"""
# Load the protein template to obtain residue information
template = gpcr_lib.load_template(mol)
residues = np.array(template.residues)
residues0 = np.arange(1, residues.shape[0] + 1)
codes = np.array(template.codes)
names = np.array(template.names)
pcontacts = []
for fi, (filename, label) in enumerate(zip(filenames, labels)):
# Read the state file from disc and perform pairwise contact analysis
states = [gpcr_lib.read_statefile(filename, every)]
pp = pycontacts.pairwise_contacts(states[-1])
# Do the same for multiple repeats and average over them
if repeats is not None:
pcontacts.append(np.zeros([len(repeats), pp.shape[0],
pp.shape[1]]))
pcontacts[-1][0, :, :] = pp
for ri, r in enumerate(repeats[1:], 1):
filename2 = filename.replace(repeats[0], r)
states.append(gpcr_lib.read_statefile(filename2, every))
pp = pycontacts.pairwise_contacts(states[-1])
pcontacts[-1][ri, :, :] = pp
pcontacts[-1] = pcontacts[-1].mean(axis=0) * 100.0
else:
pcontacts.append(pp * 100.0)
# Conversion factor from snapshot lifetime to ns lifetime
ns_per_snapshot = time / float(states[0].shape[0])
# Calculate lifetimes
all_states = np.concatenate(states, axis=1)
life_av, life_max = pycontacts.lifetime(all_states)
life_av = np.reshape(life_av, [len(repeats), residues.shape[0]]).mean(
axis=0) * ns_per_snapshot
life_max = np.reshape(life_max, [len(repeats), residues.shape[0]
]).mean(axis=0) * ns_per_snapshot
print "Median av-lifetime = %.3f" % np.median(life_av)
print "Average av-lifetime = %.3f" % np.mean(life_av)
print "Median max-lifetime = %.3f" % np.median(life_max)
print "Average max-lifetime = %.3f" % np.mean(life_max)
# Draw a 2D residue-residue joint probability plot
f2d = plt.figure(10 + fi)
_draw_2d(f2d.gca(), residues0, residues, pcontacts[-1])
f2d.savefig("%s_%s_2d.png" % (out, label), format="png")
# Draw a residue contact probability plot
f1d = plt.figure(20 + fi, figsize=(6.85, 3.41), tight_layout=True)
_draw_1d(f1d.gca(), residues0, residues, codes, template.rhelices,
pcontacts[-1].diagonal(), "Contact probability", 80)
f1d.savefig("%s_%s_1d.png" % (out, label), format="png", dpi=300)
# And print it out do disc
with open("%s_%s_1d.txt" % (out, label), "w") as f:
for (name, res, prob) in zip(names, residues,
pcontacts[-1].diagonal()):
f.write("%s%d %8.3f\n" % (name.capitalize(), res, prob))
# Draw a average lifetime plot
f1d = plt.figure(40 + fi, figsize=(6.85, 3.41), tight_layout=True)
_draw_1d(f1d.gca(), residues0, residues, codes, template.rhelices,
life_av, "Average lifetime (ns)", 0.6)
f1d.savefig("%s_%s_avlife.png" % (out, label), format="png")
# Draw a maximum lifetime plot
f1d = plt.figure(50 + fi, figsize=(6.85, 3.41), tight_layout=True)
_draw_1d(f1d.gca(), residues0, residues, codes, template.rhelices,
life_max, "Maximum lifetime (ns)", 35)
f1d.savefig("%s_%s_maxlife.png" % (out, label), format="png")
# Plot residue-type averaged occupancies
faa = plt.figure(30)
_draw_aa(faa.gca(), names, pcontacts, labels)
faa.savefig(args.out + "_aa.png", format="png")
cholmolfile = open(prefix + "_chol.mstate.%.0f.dat" % (args.cutoff),
"wb")
cholresfile = open(prefix + "_chol.resstate.%.0f.dat" % (args.cutoff),
"wb")
cholburresfile = open(
prefix + "_chol.buried.rstate.%.0f.dat" % (args.cutoff), "wb")
ohmolfile = open(prefix + "_chol-oh.mstate.%.0f.dat" % (args.cutoff),
"wb")
ohresfile = open(prefix + "_chol-oh.resstate.%.0f.dat" % (args.cutoff),
"wb")
ohburresfile = open(
prefix + "_chol-oh.buried.rstate.%.0f.dat" % (args.cutoff), "wb")
if args.reslist is not None:
with open(args.reslist, "r") as f:
lines = f.readlines()
template = gpcr_lib.load_template(lines[0].strip())
reslist = [
template.residues.index(int(l.strip())) for l in lines[1:]
]
reslistfile = open(
prefix + "_chol-oh.resliststate.%.0f.dat" % (args.cutoff),
"wb")
else:
reslist = None
reslistfile = None
if do_joint and len(chols) > 0:
jointcom = np.zeros([len(residues), len(residues)])
jointcomburr = np.zeros([len(residues), len(residues)])
jointoh = np.zeros([len(residues), len(residues)])
jointohburr = np.zeros([len(residues), len(residues)])
def _rescontacts(filenames,
labels,
repeats,
sites,
out,
mol,
time,
every,
block=None):
"""
Main analysis routine
Parameters
----------
filenames : list of strings
the group of files to analyse
labels : list of string
the label for each group
repeats : list of string
replacement pattern for multiple repeats within each group
sites : list of string
sites to group and analyse
out : string
output prefix
mol : string
protein identifier
time : float
the total simulation time
every : int
the reading frequency
block : tuple of int
just do the calculation on a block of the series
"""
# Load the protein template to obtain residue information
template = gpcr_lib.load_template(mol)
residues = np.array(template.residues)
residues0 = np.arange(1, residues.shape[0] + 1)
codes = np.array(template.codes)
names = np.array(template.names)
# Setup analysis of sites/groups
if sites:
# If the first element is l, we will do a lead-analysis
# i.e. the first residue in a site needs to be on if the whole
# site is to be considered to be on
if sites[0].lower() == "l":
makefnc = pycontacts.make_group_series_lead
sites = sites[1:]
else:
makefnc = pycontacts.make_group_series
# Find maximum member and setup group labels
maxmem = 0
grouplbl = []
for site in sites:
sidx = [template.residues.index(int(r)) for r in site.split(",")]
maxmem = max(maxmem, len(sidx))
grouplbl.append(",".join(
["%s%d" % (codes[idx], residues[idx]) for idx in sidx]))
# Fill the np array with group indices or -1
groups = np.zeros((maxmem, len(sites)), dtype=int) - 1
for i, site in enumerate(sites):
sidx = [template.residues.index(int(r)) for r in site.split(",")]
groups[:len(sidx), i] = np.asarray(sidx)
pcontacts = []
pcontacts_std = []
for fi, (filename, label) in enumerate(zip(filenames, labels)):
# Read the state file from disc and perform pairwise contact analysis
states = [gpcr_lib.read_statefile(filename, every, block)]
pp = pycontacts.pairwise_contacts(states[-1])
if sites:
gstates = [makefnc(states[-1], groups)]
pg = pycontacts.pairwise_contacts(gstates[-1]).diagonal()
# Do the same for multiple repeats and average over them
if repeats is not None:
pcontacts.append(np.zeros([len(repeats), pp.shape[0],
pp.shape[1]]))
pcontacts[-1][0, :, :] = pp
if sites:
pgroupcontacts = np.zeros([len(repeats), pg.shape[0]])
pgroupcontacts[0, :] = pg
for ri, r in enumerate(repeats[1:], 1):
filename2 = filename.replace(repeats[0], r)
states.append(gpcr_lib.read_statefile(filename2, every, block))
pp = pycontacts.pairwise_contacts(states[-1])
pcontacts[-1][ri, :, :] = pp
if sites:
gstates.append(makefnc(states[-1], groups))
pg = pycontacts.pairwise_contacts(gstates[-1]).diagonal()
pgroupcontacts[ri, :] = pg
pcontacts_std.append(pcontacts[-1].std(axis=0) * 100.0 /
np.sqrt(pcontacts[-1].shape[0]))
pcontacts[-1] = pcontacts[-1].mean(axis=0) * 100.0
if sites:
pgroupcontacts_std = pgroupcontacts.std(
axis=0) * 100.0 / np.sqrt(pgroupcontacts.shape[0])
pgroupcontacts = pgroupcontacts.mean(axis=0) * 100.0
else:
pcontacts.append(pp * 100.0)
pgroupcontacts = pg * 100
if sites:
print "Sites occupancies:"
print "\n".join([
"\t%s: %.3f +- %.3f" % (lbl, pg, std) for pg, std, lbl in zip(
pgroupcontacts, pgroupcontacts_std, grouplbl)
])
# Conversion factor from snapshot lifetime to ns lifetime
print time
if block is not None:
time = time / float(block[0]) * block[1]
print time
ns_per_snapshot = time / float(states[0].shape[0])
# Calculate lifetimes
all_states = np.concatenate(states, axis=1)
life_av, life_max = pycontacts.lifetime(all_states)
life_av = np.reshape(life_av, [len(repeats), residues.shape[0]]).mean(
axis=0) * ns_per_snapshot
life_max = np.reshape(life_max, [len(repeats), residues.shape[0]
]).mean(axis=0) * ns_per_snapshot
print "Median av-lifetime = %.3f" % np.median(life_av)
print "Average av-lifetime = %.3f" % np.mean(life_av)
print "Median max-lifetime = %.3f" % np.median(life_max)
print "Average max-lifetime = %.3f" % np.mean(life_max)
if sites:
all_gstates = np.concatenate(gstates, axis=1)
glife_av, glife_max = pycontacts.lifetime(all_gstates)
glife_av = np.reshape(glife_av, [len(repeats), groups.shape[1]
]).mean(axis=0) * ns_per_snapshot
glife_max = np.reshape(
glife_max,
[len(repeats), groups.shape[1]]).mean(axis=0) * ns_per_snapshot
print "Sites occ. av-lifetime max-lifetime:"
print "\n".join([
"\t%s: %.3f\t%.3f\t%.3f" % (lbl, o, a, m) for o, a, m, lbl in
zip(pgroupcontacts, glife_av, glife_max, grouplbl)
])
# Draw a 2D residue-residue joint probability plot
f2d = plt.figure(10 + fi)
_draw_2d(f2d.gca(), residues0, residues, pcontacts[-1])
f2d.savefig("%s_%s_2d.png" % (out, label), format="png", dpi=300)
# Draw a residue contact probability plot
f1d = plt.figure(20 + fi, figsize=(6.85, 3.41), tight_layout=True)
p = 80 if label == "com" else 100
_draw_1d(f1d.gca(), residues0, residues, codes, template.rhelices,
pcontacts[-1].diagonal(), "Contact probability", p)
f1d.savefig("%s_%s_1d.png" % (out, label), format="png", dpi=300)
# Draw a average lifetime plot
f1d = plt.figure(40 + fi, figsize=(6.85, 3.41), tight_layout=True)
_draw_1d(f1d.gca(), residues0, residues, codes, template.rhelices,
life_av, "Average lifetime (ns)", 0.6)
f1d.savefig("%s_%s_avlife.png" % (out, label), format="png", dpi=300)
# Draw a maximum lifetime plot
f1d = plt.figure(50 + fi, figsize=(6.85, 3.41), tight_layout=True)
_draw_1d(f1d.gca(), residues0, residues, codes, template.rhelices,
life_max, "Maximum lifetime (ns)", 35)
f1d.savefig("%s_%s_maxlife.png" % (out, label), format="png", dpi=300)
# And print it out do disc
with open("%s_%s_1d.txt" % (out, label), "w") as f:
for (name, res, prob, prob_std, rav,
rmax) in zip(names, residues, pcontacts[-1].diagonal(),
pcontacts_std[-1].diagonal(), life_av, life_max):
f.write("%s%d\t%8.3f\t%8.3f\t%8.3f\t%8.3f\n" %
(name.capitalize(), res, prob, prob_std, rav, rmax))
# Plot residue-type averaged occupancies
faa = plt.figure(30)
_draw_aa(faa.gca(), names, pcontacts, labels)
faa.savefig(args.out + "_aa.png", format="png", dpi=300)
shortmolfile = open(prefix+"_short.mstate.%.0f.dat"%args.cutoff,"wb")
shortresfile = open(prefix+"_short.resstate.%.0f.dat"%args.cutoff,"wb")
longmolfile = open(prefix+"_long.mstate.%.0f.dat"%args.cutoff,"wb")
longresfile = open(prefix+"_long.resstate.%.0f.dat"%args.cutoff,"wb")
if do_contacts and len(chols) > 0 :
cholmolfile = open(prefix+"_chol.mstate.%.0f.dat"%(args.cutoff),"wb")
cholresfile = open(prefix+"_chol.resstate.%.0f.dat"%(args.cutoff),"wb")
cholburresfile = open(prefix+"_chol.buried.rstate.%.0f.dat"%(args.cutoff),"wb")
ohmolfile = open(prefix+"_chol-oh.mstate.%.0f.dat"%(args.cutoff),"wb")
ohresfile = open(prefix+"_chol-oh.resstate.%.0f.dat"%(args.cutoff),"wb")
ohburresfile = open(prefix+"_chol-oh.buried.rstate.%.0f.dat"%(args.cutoff),"wb")
if args.reslist is not None:
with open(args.reslist,"r") as f :
lines = f.readlines()
template = gpcr_lib.load_template(lines[0].strip())
reslist = [template.residues.index(int(l.strip())) for l in lines[1:]]
reslistfile = open(prefix+"_chol-oh.resliststate.%.0f.dat"%(args.cutoff),"wb")
else :
reslist = None
reslistfile = None
if do_joint and len(chols) > 0 :
jointcom = np.zeros([len(residues),len(residues)])
jointcomburr = np.zeros([len(residues),len(residues)])
jointoh = np.zeros([len(residues),len(residues)])
jointohburr = np.zeros([len(residues),len(residues)])
# Allocate matrices for buried analysis
if do_countburied :
def _rescontacts(filenames,labels,repeats,out,mol,time,every) :
"""
Main analysis routine
Parameters
----------
filenames : list of strings
the group of files to analyse
labels : list of string
the label for each group
repeats : list of string
replacement pattern for multiple repeats within each group
out : string
output prefix
mol : string
protein identifier
time : float
the total simulation time
every : int
the reading frequency
"""
# Load the protein template to obtain residue information
template = gpcr_lib.load_template(mol)
residues = np.array(template.residues)
residues0 = np.arange(1,residues.shape[0]+1)
codes = np.array(template.codes)
names = np.array(template.names)
pcontacts = []
for fi,(filename,label) in enumerate(zip(filenames,labels)) :
# Read the state file from disc and perform pairwise contact analysis
states = [gpcr_lib.read_statefile(filename,every)]
pp = pycontacts.pairwise_contacts(states[-1])
# Do the same for multiple repeats and average over them
if repeats is not None :
pcontacts.append(np.zeros([len(repeats),pp.shape[0],pp.shape[1]]))
pcontacts[-1][0,:,:] = pp
for ri,r in enumerate(repeats[1:],1) :
filename2 = filename.replace(repeats[0],r)
states.append(gpcr_lib.read_statefile(filename2,every))
pp = pycontacts.pairwise_contacts(states[-1])
pcontacts[-1][ri,:,:] = pp
pcontacts[-1] = pcontacts[-1].mean(axis=0)*100.0
else :
pcontacts.append(pp*100.0)
# Conversion factor from snapshot lifetime to ns lifetime
ns_per_snapshot = time / float(states[0].shape[0])
# Calculate lifetimes
all_states = np.concatenate(states,axis=1)
life_av,life_max = pycontacts.lifetime(all_states)
life_av = np.reshape(life_av,[len(repeats),residues.shape[0]]).mean(axis=0)*ns_per_snapshot
life_max = np.reshape(life_max,[len(repeats),residues.shape[0]]).mean(axis=0)*ns_per_snapshot
print "Median av-lifetime = %.3f"%np.median(life_av)
print "Average av-lifetime = %.3f"%np.mean(life_av)
print "Median max-lifetime = %.3f"%np.median(life_max)
print "Average max-lifetime = %.3f"%np.mean(life_max)
# Draw a 2D residue-residue joint probability plot
f2d = plt.figure(10+fi)
_draw_2d(f2d.gca(),residues0,residues,pcontacts[-1])
f2d.savefig("%s_%s_2d.png"%(out,label),format="png")
# Draw a residue contact probability plot
f1d = plt.figure(20+fi,figsize=(6.85,3.41),tight_layout=True)
_draw_1d(f1d.gca(),residues0,residues,codes,template.rhelices,pcontacts[-1].diagonal(),"Contact probability",80)
f1d.savefig("%s_%s_1d.png"%(out,label),format="png",dpi=300)
# And print it out do disc
with open("%s_%s_1d.txt"%(out,label),"w") as f :
for (name,res,prob) in zip(names,residues,pcontacts[-1].diagonal()) :
f.write("%s%d %8.3f\n"%(name.capitalize(),res,prob))
# Draw a average lifetime plot
f1d = plt.figure(40+fi,figsize=(6.85,3.41),tight_layout=True)
_draw_1d(f1d.gca(),residues0,residues,codes,template.rhelices,life_av,"Average lifetime (ns)",0.6)
f1d.savefig("%s_%s_avlife.png"%(out,label),format="png")
# Draw a maximum lifetime plot
f1d = plt.figure(50+fi,figsize=(6.85,3.41),tight_layout=True)
_draw_1d(f1d.gca(),residues0,residues,codes,template.rhelices,life_max,"Maximum lifetime (ns)",35)
f1d.savefig("%s_%s_maxlife.png"%(out,label),format="png")
# Plot residue-type averaged occupancies
faa = plt.figure(30)
_draw_aa(faa.gca(),names,pcontacts,labels)
faa.savefig(args.out+"_aa.png",format="png")
def _resjointcontacts(filename, label, mat, repeats, out, mol):
"""
Main analysis routine
Parameters
----------
filename : string
file to analyse
label : string
label for the group
mat : string
matrix identifier
repeats : list of string
replacement pattern for multiple repeats
out : string
output prefix
mol : string
protein identifier
"""
# Load the protein template to obtain residue information
template = gpcr_lib.load_template(mol)
residues = np.array(template.residues)
residues0 = np.arange(1, residues.shape[0] + 1)
codes = np.array(template.codes)
names = np.array(template.names)
pcontacts = []
npz = np.load(filename)
pjoint0 = npz["joint" + mat]
pjoint = np.zeros([len(repeats), pjoint0.shape[0], pjoint0.shape[0]])
pjoint[0, :, :] = pjoint0
# Do the same for multiple repeats and average over them
if repeats is not None:
for ri, r in enumerate(repeats[1:], 1):
filename2 = filename.replace(repeats[0], r)
npz = np.load(filename2)
pjoint[ri, :, :] = npz["joint" + mat]
pjoint_std = pjoint.std(axis=0) / np.sqrt(len(repeats))
f2d = plt.figure(2, tight_layout=True)
C = gpcr_lib.draw_joint2d(f2d.gca(), residues0, residues, pjoint_std)
f2d.colorbar(C)
f2d.savefig("%s_%s_2d_std.png" % (out, label), format="png")
# Draw a 2D residue-residue joint probability plot
f2d = plt.figure(1, tight_layout=True)
C = gpcr_lib.draw_joint2d(f2d.gca(), residues0, residues,
pjoint.mean(axis=0))
f2d.colorbar(C)
f2d.gca().text(1.04, 1.01, "p(A,B)", transform=f2d.gca().transAxes)
f2d.savefig("%s_%s_2d.png" % (out, label), format="png")
f2d = plt.figure(3, tight_layout=True)
C = gpcr_lib.draw_joint2d(f2d.gca(),
residues0,
residues,
pjoint.mean(axis=0),
logit=True)
f2d.colorbar(C)
f2d.gca().text(1.02, 1.02, "ln p(A,B)", transform=f2d.gca().transAxes)
f2d.savefig("%s_%s_2d_log.png" % (out, label), format="png")