def loadadsite(self, gbestf, cutoff=5.0):
"""
gbestf is gbest.txt file path
read it and filter out all residues within cut-off
"""
gbests = re.sub("gbest.txt", "gbest_sorted.txt", gbestf)
try:
L = np.loadtxt(gbests, delimiter=",")
logger.debug("Loaded gbest: {}".format(gbests))
except:
logger.error("Can NOT read the numbers in {}".format(gbestf))
L = []
R = []
if len(L) > 0 and len(L[0]) > 1:
#The residues height was stored in second column
min_h = L[:,1].min() * 1.0
n_cutoff = min_h + cutoff
#filter out with-in cut-off
R = L[L[:,1] < n_cutoff][:,0]
logger.debug("filter cutoff={}: {}".format(n_cutoff, str(R)))
return R
def face2surface(self, resi):
"""
rotate the protein let the residue number resi face to the surface.
input argument resi should be a list which contains at least one element.
"""
#ASSUMED the list of residue (resi) is valid
self.cmd.center("protein")
proteinc = np.array(self.cmd.get_position())
xi = [1, 0, 0]
yi = [0, 1, 0]
#TODO: check resi in range
#TODO: should project the vector to x-z and y-z plane
self.cmd.center("protein and resi {}".format("+".join(
str(x) for x in resi)))
residuec = np.array(self.cmd.get_position())
rvt = residuec - proteinc
cosx = np.dot(rvt, xi) / (np.linalg.norm(rvt) * np.linalg.norm(xi))
cosy = np.dot(rvt, yi) / (np.linalg.norm(rvt) * np.linalg.norm(yi))
degx = np.degrees(np.arccos(cosx))
degy = np.degrees(np.arccos(cosy))
logger.info("Rotate the residue group to face to surface.")
logger.info("degx: {}, degy: {}".format(degx, degy))
self.protate("y", -90 - degx)
self.protate("x", -90 - degy)
self.cmd.create("protein_bak", "protein")
logger.debug(
"backup the initial protein position after face to surface")
#fpath = os.path.join(self.args.outdir, "bak_init.pdb")
#self.cmd.save(fpath, "protein surface")
#logger.debug("backup the initial orientation pdb file: {}".format(fpath))
#Compute the radius
#cosine = adjacent over hypotenuse
hypoten = 0
for i in resi:
self.cmd.center("protein and resi {}".format(str(i)))
ric = np.array(self.cmd.get_position())
length = np.linalg.norm(ric - proteinc)
if length > hypoten:
hypoten = length
adjacent = np.linalg.norm(rvt)
cosTheta = adjacent / (hypoten + 5
) #hard code! 5 is the adsorption site cut-off
#It should even works if only one residue in the list
Theta = np.degrees(np.arccos(cosTheta))
logger.debug("Adjacent({}) over hypoten({}) +5".format(
adjacent, hypoten))
logger.debug("Computed cosTheta: {}, and Theta: {}".format(
cosTheta, Theta))
return Theta
def ptransl(self, axis, step):
if self.pcheck(axis, step, "ptransl"): return -1
self.cmd.center("protein")
center = self.cmd.get_position()
logger.debug(
"[ptransl] protein position BEFORE translation: {}".format(center))
delta = [0.0, 0.0, 0.0]
delta[self.xyzid[axis]] = step
logger.debug("[ptransl] translate protein delta: {}".format(delta))
self.cmd.translate(delta, "protein", state=0, camera=0)
self.cmd.center("protein")
center = self.cmd.get_position()
logger.debug(
"[ptransl] protein position AFTER translation: {}".format(center))
def scoring(self):
logger.info("Now we call gromacs to calculate the score.")
for con in self.jsdb["confs"]:
f = con["id"] + ".pdb"
logger.debug("processing file: {}".format(f))
fn = os.path.join(self.jsdb["confd"], f)
logger.debug("scoringone the file: {}".format(fn))
r = self.scoringone(fn)
try:
con["stepN"] = r["stepN"]
con["coul"] = r["coul"]
con["ljsr"] = r["ljsr"]
con["file"] = r["file"]
logger.debug("updated the configuration db, {}:{}".format(
con["coul"], con["ljsr"]))
except:
logger.warning("scoringone return incorrect data!!!")
self.savejdb()
def initcmd(self):
logger.info("Initialized command line arguments")
self.parser = argparse.ArgumentParser(
description="This is PSO testing")
self.parser.add_argument(
"--outdir",
help="set the output configuration directory.",
required=False,
default="pso_conf")
self.parser.add_argument("--jsdbf",
help="set the json db file path.",
default="db.json")
self.parser.add_argument("--proteinf",
help="set the input protein pdb filename.",
default="protein.pdb",
metavar="protein.pdb")
self.parser.add_argument("--surfacef",
help="set the input surface pdb filename.",
default="surface.pdb",
metavar="surface.pdb")
self.parser.add_argument("--n",
help="set the number of birds.",
default=200,
required=False,
type=int)
self.parser.add_argument("--r",
help="set the total iteration number.",
default=10,
required=False,
type=int)
self.parser.add_argument("--w",
help="set the weight for updating velocity.",
default=0.721,
required=False,
type=float)
self.parser.add_argument("--c1",
help="set the parameter C1.",
default=1.193,
required=False,
type=float)
self.parser.add_argument("--c2",
help="set the parameter C2.",
default=1.193,
required=False,
type=float)
self.parser.add_argument("--maxitr",
help="set the maximum number of iteration.",
default=0,
required=False,
type=int)
self.parser.add_argument(
"--emdir",
help="set the directory for EM, energy minimization using Gromacs.",
default="EM",
required=False,
type=str)
self.parser.add_argument(
"--keep-pdb",
help="keep all of the pdf files generated during the searching.",
default=False,
action="store_true",
required=False)
self.initpg = self.parser.add_mutually_exclusive_group()
self.initpg.add_argument(
"--resi",
help="searching only the prefered residue surface",
type=int,
nargs="+")
self.initpg.add_argument("--offset",
help="set the initial searching orientation",
nargs=6,
type=float)
self.initpg.add_argument(
"--init",
help="use the exactly same molecule conformation as input",
action="store_true",
default=False)
self.args = self.parser.parse_args()
logger.debug("args: {}".format(self.args))
def initvar(self):
super(simplePSO, self).initvar()
logger.debug("initialized PSO variables: {}".format(self.__dict__))
self.tid = 1
def plot(self):
'''output the db.json to csv'''
if self.args.no_plot:
logger.info("skipped the plot progress...")
return 0
try:
a = self.jsdb["confs"]
steps = self.jsdb["steps"]
mindst = self.jsdb["mindst"]
except KeyError:
logger.error(
"Can not find the confs/steps from the json db file. It might be corrupted."
)
return 1
count = 0
x = []
y = []
z = []
xt = []
yt = []
zt = []
ang = 360.0 / steps
lx = []
ly = []
ld = []
minres = "UNKNOW"
mineng = 7777777
mincnt = 0
maxeng = -7777777
maxres = "UNKNOW"
minxra = 0
minyra = 0
logger.debug("preparing X Y Z")
for i in a:
try:
enegy = float(i["coul"]) + float(i["ljsr"])
resnm = i["id"]
xi = int(i["xi"])
yi = int(i["yi"])
except (ValueError, KeyError):
enegy = 7777777
resnm = "UNKNOW"
xi = 0
yi = 0
if enegy < mineng:
mineng = enegy
minres = resnm
mincnt = count
minxra = xi
minyra = yi
logger.debug("Found a lower energy residue: {}@{}".format(
minres, mineng))
if enegy > maxeng:
maxeng = enegy
logger.debug("Found a higher energy residue: {}@{}".format(
maxres, maxeng))
xti = float(xi * ang - self.args.init_xr)
yti = float(yi * ang - self.args.init_yr)
xt.append(xti)
yt.append(yti)
zt.append(enegy)
lx.append(count)
ly.append(enegy)
ld.append(resnm)
count += 1
if (count % steps) == 0:
x.append(xt)
xt = []
y.append(yt)
yt = []
z.append(zt)
zt = []
x = np.array(x)
y = np.array(y)
z = np.array(z)
logger.debug("get X, Y and Z, then plot the graph.")
figa = plt.figure()
cmap = plt.cm.get_cmap("RdBu_r")
if self.args.minlv != None and self.args.maxlv != None:
a = self.args.minlv
b = self.args.maxlv
i = (
b - a
) / self.ctlvs #ASSUMED: the maxlv is always bigger than minlv
levels = range(a, b, i)
logger.debug("The contour map levels: {}".format(levels))
else:
levels = None
logger.debug("Use the default contour map levels.")
plt.plot(112, -65, 'ko') #the PSO ans
#plot the minimum location
plt.plot(float(minxra * ang - self.args.init_xr),
float(minyra * ang - self.args.init_yr), 'kx')
Da = plt.contourf(x,
y,
z,
cmap=cmap,
levels=levels,
norm=mpl.colors.SymLogNorm(011))
plt.title(u"contour diagram\ndistance={}Å".format(mindst))
plt.xlabel("X rotation angle")
plt.ylabel("Y rotation angle")
cbar = plt.colorbar(Da)
cbar.ax.set_ylabel("energy level")
plt.savefig("diagram_0a.pdf")
logger.debug("plot contour diagram and save as pdf file.")
figb = plt.figure()
plt.title(u"energy line\ndistance={}Å".format(mindst))
plt.xlabel("Iteration Number")
plt.ylabel("Energy Value")
Db = plt.plot(lx, ly, 'k')
plt.plot(mincnt, mineng, 'bo')
plt.plot([mincnt, mincnt * 1.1], [mineng, mineng], 'k')
plt.text(mincnt * 1.13,
mineng,
"id: {}\nen: {}".format(minres, mineng),
verticalalignment="center",
horizontalalignment="left")
logger.debug("plot energy line only.")
if self.args.minlv != None:
plt.ylim(ymin=self.args.minlv)
logger.debug("set the y-axis minimum range.")
if self.args.maxlv != None:
plt.ylim(ymax=self.args.maxlv)
logger.debug("set the y-axis maximum range.")
plt.savefig("diagram_0k.pdf")
ly = np.array(ly)
ld = np.array(ld)
L = np.column_stack((ly, ld))
np.savetxt("energy.txt.gz", L, delimiter=" ", fmt="%11s %11s")
logger.debug("plot energy line diagram and save as pdf file.")
fige = plt.figure()
plt.title(u"normalized energy line\ndistance={}Å".format(mindst))
plt.ylabel("Energy Value")
plt.xlabel("residues")
plt.axis("off")
plt.grid("on")
plt.xticks([])
plt.yticks([])
sly = np.sort(ly)
sry = sly[::-1]
nly = (sry - mineng) / (maxeng - mineng)
De = plt.plot(range(len(nly)), nly)
plt.text(0, 0, "id: {}@{}".format(minres, mineng))
logger.debug("plot energy histogram.")
plt.savefig("diagram_0e.pdf")
logger.debug("plot energy histogram diagram and saved as pdf file.")
figh = plt.figure()
plt.title(u"normalized energy line\ndistance={}Å".format(mindst))
Dh = plt.hist(ly, 100)
plt.xlabel("The lowest configuration is id: {}@eng: {}".format(
minres, mineng))
plt.savefig("diagram_0h.pdf")
logger.debug("plot another histogram diagram and saved as pdf file.")
if self.args.no_plot_re:
logger.info("skipped the residues plot progress...")
return 0
logger.debug(
"plot residues configuration graph, go through all configurations..."
)
self.cmd.load("system00003.pse")
for con in a:
cid = con["id"]
try:
nid = re.findall('\d+', cid)[0]
nid = int(nid)
except:
logger.warning("function plot - can not find digit from cid.")
nid = 0
logger.debug("find the digit: {} from cid.".format(nid))
cen = float(con["coul"]) + float(con["ljsr"])
logger.debug("processing conf: {}, the energy value is: {}".format(
cid, cen))
logger.debug("Now create the system state for processing...")
self.cmd.create("system", "mov", nid + 1, 1)
atoms = self.cmd.get_model("system")
resds = atoms.get_residues()
xr = [] # x axis: residues id
yr = [] # y axis: mini distance
mindt = [999999.99] * len(resds)
surfc = self.surfaceh()
szv = surfc["stop"]
for atom in atoms.atom:
rid = int(atom.resi) - 1
azv = atom.coord[2]
dst = azv - szv
if dst < mindt[rid]:
mindt[rid] = dst
#save the graph under jsdb["confd"] directory
figs = plt.figure()
Ds = plt.plot(range(1, len(mindt)+1), mindt, 'r+', \
range(1, len(mindt)+1), mindt, 'k')
plt.title(
u"residues configuraions diagram\ndistance={}Å; energy={}kj".
format(self.jsdb["mindst"], cen))
plt.savefig(os.path.join(self.jsdb["confd"], cid + ".pdf"))
logger.debug("plot a residues diagram and saved as pdf.")
def protation(self):
s = self.args.steps
step = 360.0 / s
sum = 1
for i in range(s):
for j in range(s):
#save state
self.cmd.create("mov", "protein", self.cmd.get_state(), sum)
logger.debug(
"[Current state ({})]Saved new state (count: {}), after rotate x"
.format(self.cmd.get_state(), i * s + j + 1))
logger.debug(
"[protation] for loop i: {}, j: {}, ia: {}, ja: {}".format(
i, j, i * step, j * step))
logger.debug("{:=^70}".format("next round"))
self.protate("x", step)
sum += 1
logger.debug("{:=^70}".format("Y AXIS"))
self.protate("y", step)
logger.debug("{:=^70}".format("N TURN"))
self.cmd.ending()
logger.debug("[protation] Goes to the last state ({}).".format(
self.cmd.get_state()))
self.cmd.create("final", "protein", self.cmd.get_state())
logger.debug("Saved the final structure.")
self.savefile(pdb=True)
logger.info(
"[protation] Saved file (count: {}), rotation sampling is done".
format(self.svcount))
def protate(self, axis, step):
if self.pcheck(axis, step, "protate"): return -1
za = self.proteinz()
logger.debug("[protate] The lowest Z B4 rotate is : {}".format(za))
self.cmd.center("protein")
origin = self.cmd.get_position()
logger.debug("[protate] define the origin: {}".format(origin))
self.cmd.rotate(axis, step, "protein", camera=0, origin=origin)
logger.debug("[protate] Rotate {} axis {} units ".format(axis, step))
logger.debug("rotate {}, {}, protein, camera=0, origin={}".format(
axis, step, origin))
zb = self.proteinz()
logger.debug("[protate] The lowest Z AF rotate is : {}".format(zb))
delta = za - zb
logger.debug("[protate] distA: {} - distB: {} = delta: {}".format(
za, zb, delta))
self.cmd.translate([0, 0, delta], "protein", state=0, camera=0)
logger.debug(
"[protate] translated the protein {}(delta) units.".format(delta))
logger.debug("[protate] now the Z value is : {}".format(
self.proteinz()))
def mindst(self, state=-1, z=True, surfc=None):
'''Find the minimal distance bewteen the surface and the protein
if the mov model not yet exist, please give state=0 to choose the protein model
z=False may not work in current version'''
pdown = float("+inf")
adown = [0.0, 0.0, 0.0]
rdown = "" #resi of atom
ndown = "" #name of atom
if surfc:
surf = surfc
else:
surf = self.surfaceh()
stop = surf["stop"]
atop = surf["atop"]
rtop = surf["rtop"]
ntop = surf["ntop"]
#loop all atoms to find the top of surface and the bottom of protein
try:
int(state)
except ValueError:
logger.warning("given state is not integer!!")
return -1
#TODO: extract surface height to another function
try:
atoms = self.cmd.get_model("surface")
except:
logger.warning("Can not find model: surface")
return 0
for atom in atoms.atom:
if atom.coord[2] > stop:
stop = atom.coord[2]
atop = atom.coord[:3]
rtop = str(atom.resi)
ntop = str(atom.name)
logger.debug("(state: {})sTOP is : {}.".format(self.cmd.get_state(),
stop))
model = ""
get_state = self.cmd.get_state()
if state == 0:
model = "protein"
get_state = 0
elif state != -1:
model = "mov"
get_state = state
else:
model = "mov"
try:
logger.debug("Trying to get model: {} at state({}).".format(
model, get_state))
atoms = self.cmd.get_model(model, get_state)
except:
logger.warning("Can not find model({}): in state({})".format(
model, get_state))
return 0
for atom in atoms.atom:
if atom.coord[2] < pdown:
pdown = atom.coord[2]
adown = atom.coord[:3]
rdown = str(atom.resi)
ndown = str(atom.name)
logger.debug("(state: {})pdown is : {}.".format(
self.cmd.get_state(), pdown))
logger.debug("Calculate the destince bewteen two atoms.")
deltasq = [(i - j)**2 for i, j in zip(atop, adown)]
distanc = math.sqrt(sum(deltasq))
distanz = pdown - stop
logger.debug("The distance is : {}".format(distanc))
logger.debug("Z distance is : {}".format(distanz))
r = distanc
if z == True:
r = distanz
return r
def load_post(self, skip=False):
#ASSUMED: it is called after load_file
if not skip:
self.cmd.center("surface")
surfacec = self.cmd.get_position()
logger.debug("Center of the surface: " +
", ".join(format(x, "f") for x in surfacec))
self.cmd.center("protein")
proteinc = self.cmd.get_position()
logger.debug("Center of the protein: " +
", ".join(format(x, "f") for x in proteinc))
delta = [i - j for i, j in zip(proteinc, surfacec)]
logger.debug("move delta: " +
", ".join(format(x, "f") for x in delta))
self.cmd.translate(delta, "surface", camera=0)
logger.debug("Translate Z axis of protein-surface.")
miniZ = self.mindst(state=0, z=True)
self.cmd.translate(
[0, 0, miniZ - 0], "surface",
camera=0) #let the distance to be zero, translate later
logger.debug("The minimal Z distance now is : {}".format(
self.mindst(state=0, z=True)))
#logger.info("The initial structure has been created ")
self.cmd.create("protein_bak", "protein")
logger.debug("backup the original initial protein position")
def load_file(self):
logger.debug("loading " + self.args.proteinf)
self.cmd.load(self.args.proteinf, "protein")
logger.debug("loading " + self.args.surfacef)
self.cmd.load(self.args.surfacef, "surface")
logger.info("loaded protein and surface pdb files.")
def cluster(self):
"""
cluster by dbscae
it will return a list with the corresponding cluster id
we convert it into a dict index by cluster id
"""
self.get_gbest()
X = StandardScaler().fit_transform(self.Hnums)
db = DBSCAN(eps=self.args.eps, min_samples=self.args.minn).fit(X)
self.dbresult = np.array(db.labels_)
logger.debug("DBSCAN labels: {}".format(",".join(map(str,self.dbresult))))
"""
self.clusters = [ [] for x in range(labels.max()+2) ]
for idx, val in enumerate(labels):
if idx < len(self.labels) and val < len(self.clusters):
self.clusters[val].append(self.labels[idx])
self.ranking()
if not self.args.no_plot and len(self.clusters) > 0:
idx = 0
while idx < len(self.rankings) and idx < len(self.rankinge):
rankids = self.rankings[idx]
rankide = self.rankinge[idx]
loweids = self.low_energy[rankids]
loweide = self.low_energy[rankide]
s_clusters = self.clusters[rankids]
e_clusters = self.clusters[rankide]
#self.hello("DEBUG the lowest energy by size is {}".format(s_clusters[loweids]))
#self.hello("DEBUG the lowest energy by energy is {}".format(e_clusters[loweide]))
#self.plotcluster(idx+1, s_clusters, "bysize", s_clusters[loweids])
self.plotcluster("{:02d}".format(idx+1), e_clusters, "byenergy", e_clusters[loweide])
idx+=1
self.hello("Plot the noise group")
self.plotcluster("noise", self.clusters[-1], "noise")
"""
for idx, cid in enumerate(self.dbresult):
#Here, idx is *gbest_id*
if cid not in self.clusters:
self.clusters[cid] = {}
self.clusters[cid]["coden"] = "{:02d}".format(cid+1)
self.clusters[cid]["gbest"] = []
self.clusters[cid]["englt"] = []
self.clusters[cid]["adsit"] = []
self.clusters[cid]["adslt"] = []
if cid == -1:
self.clusters[cid]["coden"] = "noise"
gbestf = self.labels[idx]
energy = self.loadenergy(gbestf)
adsita = self.loadadsite(gbestf)
adsitb = self.clusters[cid]["adsit"]
logger.debug("site a:{}".format(adsita))
logger.debug("site b:{}".format(adsitb))
if len(adsitb) > 0:
adsitc = np.intersect1d(adsita, adsitb)
logger.debug("site A and B intersection.")
else:
adsitc = adsita
logger.debug("site c:{}".format(adsitc))
self.clusters[cid]["adslt"] += list(adsita)
self.clusters[cid]["gbest"].append(idx)
self.clusters[cid]["englt"].append(energy)
self.clusters[cid]["adsit"] = adsitc
#ASSUMED: lambda should correct
self.cs_eid = sorted(self.clusters, key=lambda x: np.mean(self.clusters[x]["englt"]), reverse=False)
if -1 in self.cs_eid:
self.cs_eid.remove(-1)
self.cs_eid.append(-1)
logger.debug("cluster sorted by energy: {}".format(str(self.cs_eid)))
for i, c in enumerate(self.cs_eid):
if c != -1:
self.clusters[c]["coden"] = "{:02d}".format(i+1)
if not self.args.no_plot and len(self.clusters) > 0:
cidlt = self.clusters.keys()
for c in cidlt:
list_of_gbesti = self.clusters[c]["gbest"]
list_of_gbestf = [ self.labels[x] for x in list_of_gbesti ]
logger.debug("gbesti: {}".format(str(list_of_gbesti)))
logger.debug("gbestf: {}".format(str(list_of_gbestf)))
filename = "{}".format(self.clusters[c]["coden"])
fpath = self.plotcluster(filename, list_of_gbestf, bold="")
self.clusters[c]["fpath"] = fpath
logger.debug("clusters: \n{}".format(self.pp.pformat(self.clusters)))
"""
for idx, cst in enumerate(self.clusters):
self.hello("DEBUG cluster id {:02d} average energy {} lowest energy {} size {}".format(idx, self.avg_energy[idx], self.lowest_eng[idx], len(cst)))
self.hello("{}".format(cst))
self.clusterbar()
self.hello("{:=^70}".format("FINALLY"))
self.hello("{:=^70}".format("by size"))
for idx, cst in enumerate(self.rankings):
cstlst = self.clusters[cst]
loweng = self.low_energy[cst]
#self.hello("The {:02d}th cluster (size={}): {}".format(idx+1, len(self.clusters[cst]), self.clusters[cst]))
self.hello("The {:02d}th cluster (size={}; average energy={}; median energy={}; lowest energy={}; energy std={}): \n{}".format(\
idx+1, len(cstlst), self.avg_energy[cst], self.median_eng[cst], self.lowest_eng[cst], self.std_energy[cst], cstlst[loweng]))
self.hello("{:=^70}".format("by energy"))
for idx, cst in enumerate(self.rankinge):
cstlst = self.clusters[cst]
loweng = self.low_energy[cst]
#self.hello("The {:02d}th cluster (energy={}): {}".format(idx+1, self.avg_energy[cst], self.clusters[cst]))
self.hello("The {:02d}th cluster (size={}; energy={}): {}".format(idx+1, len(cstlst), self.avg_energy[cst], cstlst[loweng]))
"""
self.printoutcluster()
def hello(self, msg="はじまるよ～♪ "):
logger.debug("[hello] {}".format(msg))
logger.debug("hello function is deprecated.")
self.hello("{}: {}".format(i, act.__doc__))
self.hello("="*77)
def main(self):
logger.debug("command line arguments:")
logger.debug(str(self.args))
act = getattr(self, self.args.act, self.helping)
if callable(act):
logger.debug("Run the action ({})".format(str(act)))
act()
if __name__ == "__main__":
m=dbcluster()
logger.debug("Hello! This is main function of dbcluster")
m.hello()
#m.fplot()
#m.fneo()
#m.fthree()
#m.ffour()
#m.ffive()
m.main()