def RTTplot_high_res(f,grid_cell_size=1.,burnin=0,max_age=0):
wd = "%s" % os.path.dirname(f)
name_file=os.path.splitext(os.path.basename(f))[0]
t=loadtxt(f, skiprows=max(1,burnin))
head = next(open(f)).split() # should be faster
sp_ind = [head.index(s) for s in head if "lambda_" in s]
ex_ind = [head.index(s) for s in head if "mu_" in s]
root_ind = head.index("root_age")
death_ind = head.index("death_age")
sp_shift_ind = [head.index(s) for s in head if "shift_sp_" in s]
ex_shift_ind = [head.index(s) for s in head if "shift_ex_" in s]
min_root_age = min(t[:,root_ind])
if max_age> 0: min_root_age=max_age
max_death_age= max(t[:,death_ind])
n_bins= int((min_root_age-max_death_age)/grid_cell_size)
grid = np.linspace(min_root_age,max_death_age,n_bins)
n_samples = np.shape(t)[0]
m_sp_matrix = np.zeros((n_samples,n_bins))
m_ex_matrix = np.zeros((n_samples,n_bins))
print "Extracting marginal rates..."
for i in range(n_samples):
l_shift_times = np.array([min_root_age]+ list(t[i,sp_shift_ind])+[max_death_age])
l_rates = t[i,sp_ind]
m_sp_matrix[i] = get_marginal_rates(l_shift_times,l_rates,grid)
m_shift_times = np.array([min_root_age]+ list(t[i,ex_shift_ind])+[max_death_age])
m_rates = t[i,ex_ind]
m_ex_matrix[i] = get_marginal_rates(m_shift_times,m_rates,grid)
res = np.zeros((n_bins,7)) # times, l, lM, lm, m, mM, mm
print "Calculating HPDs..."
for i in range(n_bins):
l_HPD = list(util.calcHPD(m_sp_matrix[:,i],0.95))
l_mean = mean(m_sp_matrix[:,i])
m_HPD = list(util.calcHPD(m_ex_matrix[:,i],0.95))
m_mean = mean(m_ex_matrix[:,i])
res[i] = np.array([grid[i],l_mean]+l_HPD+[m_mean]+m_HPD)
Rfile=r_plot_code(res,wd,name_file)
out="%s/%s_RTT.r" % (wd,name_file)
newfile = open(out, "w")
newfile.writelines(Rfile)
newfile.close()
print "\nAn R script with the source for the RTT plot was saved as: %sRTT.r\n(in %s)" % (name_file, wd)
if platform.system() == "Windows" or platform.system() == "Microsoft":
cmd="cd %s; Rscript %s\%s_RTT.r" % (wd,wd,name_file)
else:
cmd="cd %s; Rscript %s/%s_RTT.r" % (wd,wd,name_file)
os.system(cmd)
print "done\n"
# args
#burnin=0
#grid_cell_size=.1 # in Myr (approximated)
#f= "/Users/daniele/Desktop/try/catalina_temp/par_est/combined_10mcmc_files.log"
#max_age=23.03
#RTTplot_high_res(f,grid_cell_size,burnin,max_age)
def get_marginal_rates(f_name, min_age, max_age, nbins=0, burnin=0.2):
# returns a list of 5 items:
# 1. a vector of times (age of each marginal rate)
# 2-4. mean, min and max marginal rates (95% HPD)
# 5. a vector of times of rate shift
f = open(f_name, 'U')
if nbins == 0:
nbins = int(max_age - 0)
post_rate = f.readlines()
bins_histogram = np.linspace(min_age, max_age, nbins + 1)
marginal_rates_list = []
times_of_shift = []
if burnin < 1: # define burnin as a fraction
burnin = min(int(burnin * len(post_rate)), int(0.9 * len(post_rate)))
else:
burnin = int(burnin)
for i in range(burnin, len(post_rate)):
row = np.array(post_rate[i].split()).astype(float)
if len(row) == 0:
continue
elif len(row) == 1:
marginal_rates = np.zeros(nbins) + row[0]
else:
ind_rates = np.arange(0, int(np.ceil(len(row) / 2.)))
ind_shifts = np.arange(int(np.ceil(len(row) / 2.)), len(row))
rates = row[ind_rates]
shifts = row[ind_shifts]
#shifts = shifts[shifts>min_age]
h = np.histogram(row[ind_shifts], bins=bins_histogram)[0][::-1]
marginal_rates = rates[np.cumsum(h)]
times_of_shift += list(shifts)
marginal_rates_list.append(marginal_rates)
marginal_rates_list = np.array(marginal_rates_list)
mean_rates = np.median(marginal_rates_list, axis=0)
min_rates, max_rates = [], []
for i in range(nbins):
hpd = util.calcHPD(marginal_rates_list[:, i], 0.95)
min_rates += [hpd[0]]
max_rates += [hpd[1]]
time_frames = bins_histogram - abs(bins_histogram[1] -
bins_histogram[0]) / 2.
#print time_frames, min(times_of_shift), min_age
#quit()
time_frames = time_frames[1:]
#print len(time_frames),len(mean_rates),
n_mcmc_samples = len(
post_rate
) - burnin # number of samples used to normalize frequencies of rate shifts
return [
time_frames, mean_rates,
np.array(min_rates),
np.array(max_rates),
np.array(times_of_shift), n_mcmc_samples, marginal_rates_list
]
def plot_net_rate(resS,resE,col,min_age,max_age,plot_title,n_bins):
#computes and plots net RATES
marginal_rates_list = resS[6]-resE[6]
mean_rates= np.mean(marginal_rates_list,axis=0)
min_rates,max_rates=[],[]
for i in range(n_bins):
hpd = util.calcHPD(marginal_rates_list[:,i],0.95)
min_rates += [hpd[0]]
max_rates += [hpd[1]]
out_str = "\n#Net Diversification Rate"
out_str += util.print_R_vec("\ntime",resS[0]-min_age)
minXaxis,maxXaxis= max_age-min_age,min_age-min_age
#right now I don't have support for log, but I think this is less likely to be needed for net rates
out_str += util.print_R_vec("\nnet_rate",mean_rates[::-1])
out_str += util.print_R_vec("\nnet_minHPD",np.array(min_rates[::-1]))
out_str += util.print_R_vec("\nnet_maxHPD",np.array(max_rates[::-1]))
out_str += "\nplot(time,time,type = 'n', ylim = c(%s, %s), xlim = c(%s,%s), ylab = 'Net Rate', xlab = 'Time',lwd=2, main='%s', col= '%s' )" \
% (min(0,1.1*np.nanmin(min_rates)),1.1*np.nanmax(max_rates),minXaxis,maxXaxis,plot_title,col)
out_str += "\npolygon(c(time, rev(time)), c(net_maxHPD, rev(net_minHPD)), col = alpha('%s',0.3), border = NA)" % (col)
out_str += "\nlines(time,net_rate, col = '%s', lwd=2)" % (col)
out_str += "\nabline(h=0,lty=2)\n"
return out_str
def plot_net_rate(resS, resE, col, min_age, max_age, plot_title, n_bins):
#computes and plots net RATES
marginal_rates_list = resS[6] - resE[6]
mean_rates = np.mean(marginal_rates_list, axis=0)
min_rates, max_rates = [], []
for i in range(n_bins):
hpd = util.calcHPD(marginal_rates_list[:, i], 0.95)
min_rates += [hpd[0]]
max_rates += [hpd[1]]
out_str = "\n#Net Diversification Rate"
out_str += util.print_R_vec("\ntime", resS[0] - min_age)
minXaxis, maxXaxis = max_age - min_age, min_age - min_age
#right now I don't have support for log, but I think this is less likely to be needed for net rates
out_str += util.print_R_vec("\nnet_rate", mean_rates[::-1])
out_str += util.print_R_vec("\nnet_minHPD", np.array(min_rates[::-1]))
out_str += util.print_R_vec("\nnet_maxHPD", np.array(max_rates[::-1]))
out_str += "\nplot(time,time,type = 'n', ylim = c(%s, %s), xlim = c(%s,%s), ylab = 'Net Rate', xlab = 'Time',lwd=2, main='%s', col= '%s' )" \
% (min(0,1.1*np.nanmin(min_rates)),1.1*np.nanmax(max_rates),minXaxis,maxXaxis,plot_title,col)
out_str += "\npolygon(c(time, rev(time)), c(net_maxHPD, rev(net_minHPD)), col = alpha('%s',0.3), border = NA)" % (
col)
out_str += "\nlines(time,net_rate, col = '%s', lwd=2)" % (col)
out_str += "\nabline(h=0,lty=2)\n"
return out_str
def get_marginal_rates(f_name,min_age,max_age,nbins=0,burnin=0.2): # returns a list of 5 items: # 1. a vector of times (age of each marginal rate) # 2-4. mean, min and max marginal rates (95% HPD) # 5. a vector of times of rate shift f = file(f_name,'U') if nbins==0: nbins = int(max_age-0) post_rate=f.readlines() bins_histogram = np.linspace(0,max_age,nbins+1) marginal_rates_list = [] times_of_shift = [] if burnin<1: # define burnin as a fraction burnin=min(int(burnin*len(post_rate)),int(0.9*len(post_rate))) else: burnin = int(burnin) for i in range(burnin,len(post_rate)): row = np.array(post_rate[i].split()).astype(float) if len(row)==0: continue elif len(row)==1: marginal_rates = np.zeros(nbins)+row[0] else: ind_rates = np.arange(0,int(np.ceil(len(row)/2.))) ind_shifts = np.arange(int(np.ceil(len(row)/2.)),len(row)) rates = row[ind_rates] shifts = row[ind_shifts] #shifts = shifts[shifts>min_age] h = np.histogram(row[ind_shifts],bins =bins_histogram)[0][::-1] marginal_rates = rates[np.cumsum(h)] times_of_shift += list(shifts) marginal_rates_list.append(marginal_rates) marginal_rates_list = np.array(marginal_rates_list) mean_rates= np.median(marginal_rates_list,axis=0) min_rates,max_rates=[],[] for i in range(nbins): hpd = util.calcHPD(marginal_rates_list[:,i],0.95) min_rates += [hpd[0]] max_rates += [hpd[1]] time_frames = bins_histogram-bins_histogram[1]/2. #print time_frames, min(times_of_shift), min_age #quit() time_frames = time_frames[1:] #print len(time_frames),len(mean_rates), n_mcmc_samples = len(post_rate)-burnin # number of samples used to normalize frequencies of rate shifts return [time_frames,mean_rates,np.array(min_rates),np.array(max_rates),np.array(times_of_shift),n_mcmc_samples, marginal_rates_list]
def plot_net_rate(resS, resE, col, min_age, max_age, plot_title, n_bins):
#computes and plots net RATES
resS_marginal_rate = resS[6]
resE_marginal_rate = resE[6]
# in case they have different number of samples
max_indx = np.min(
[resS_marginal_rate.shape[0], resE_marginal_rate.shape[0]])
marginal_rates_list = resS_marginal_rate[
0:max_indx, :] - resE_marginal_rate[0:max_indx, :]
mean_rates = np.mean(marginal_rates_list, axis=0)
min_rates, max_rates = [], []
time_ax = []
for i in range(n_bins):
hpd = util.calcHPD(marginal_rates_list[:, i], 0.95)
min_rates += [hpd[0]]
max_rates += [hpd[1]]
times = abs(resS[0])
indx = np.intersect1d((times >= min_age).nonzero()[0],
(times <= max_age).nonzero()[0])
times = times[indx]
out_str = "\n#Net Diversification Rate"
out_str += util.print_R_vec("\ntime", -times)
minXaxis, maxXaxis = min_age, max_age
mean_rates = np.array(mean_rates)[::-1]
min_rates = np.array(min_rates)[::-1]
max_rates = np.array(max_rates)[::-1]
mean_rates = mean_rates[indx]
min_rates = min_rates[indx]
max_rates = max_rates[indx]
out_str += util.print_R_vec("\nnet_rate", mean_rates)
out_str += util.print_R_vec("\nnet_minHPD", np.array(min_rates))
out_str += util.print_R_vec("\nnet_maxHPD", np.array(max_rates))
out_str += "\nplot(time,time,type = 'n', ylim = c(%s, %s), xlim = c(%s,%s), ylab = 'Net Rate', xlab = 'Time',lwd=2, main='%s', col= '%s' )" \
% (min(0,1.1*np.nanmin(min_rates)),1.1*np.nanmax(max_rates),-maxXaxis,-minXaxis,plot_title,col)
out_str += "\npolygon(c(time, rev(time)), c(net_maxHPD, rev(net_minHPD)), col = alpha('%s',0.3), border = NA)" % (
col)
out_str += "\nlines(time,net_rate, col = '%s', lwd=2)" % (col)
out_str += "\nabline(h=0,lty=2)\n"
return out_str
Temp_at_events, shift_ind,
root_age)
marginal_L.append(l_vec)
marginal_M.append(m_vec)
marginal_L = np.array(marginal_L)
marginal_M = np.array(marginal_M)
l_vec = np.zeros(np.shape(marginal_L)[1])
m_vec = np.zeros(np.shape(marginal_L)[1])
hpd_array_L = np.zeros((2, np.shape(marginal_L)[1]))
hpd_array_M = np.zeros((2, np.shape(marginal_L)[1]))
for i in range(np.shape(marginal_L)[1]):
l_vec[i] = np.mean(marginal_L[:, i])
m_vec[i] = np.mean(marginal_M[:, i])
hpd_array_L[:, i] = calcHPD(marginal_L[:, i])
hpd_array_M[:, i] = calcHPD(marginal_M[:, i])
print "done"
# write R file
print "\ngenerating R file...",
out = "%s/%s_RTT.r" % (output_wd, name_file)
newfile = open(out, "wb")
if platform.system() == "Windows" or platform.system() == "Microsoft":
wd_forward = os.path.abspath(output_wd).replace('\\', '/')
r_script = "\n\npdf(file='%s/%s_RTT.pdf',width=0.6*20, height=0.6*20)\nlibrary(scales)\n" % (
wd_forward, name_file)
else:
r_script = "\n\npdf(file='%s/%s_RTT.pdf',width=0.6*20, height=0.6*20)\nlibrary(scales)\n" % (
output_wd, name_file)
r_script += print_R_vec("\n\nt", age_vec)
l_vec = np.zeros(np.shape(marginal_L)[1])
m_vec = np.zeros(np.shape(marginal_L)[1])
hpd_array_L = np.zeros((2, np.shape(marginal_L)[1]))
hpd_array_M = np.zeros((2, np.shape(marginal_L)[1]))
hpd_array_L50 = np.zeros((2, np.shape(marginal_L)[1]))
hpd_array_M50 = np.zeros((2, np.shape(marginal_L)[1]))
if i >= 0:
l_vec = np.mean(marginal_L, axis=0) # get_mode
m_vec = np.mean(marginal_M, axis=0) # get_mode
else:
for ii in range(
np.shape(marginal_L)[1]): # loop over marginal rates
l_vec[ii] = np.mean(marginal_L[:, ii]) # get_mode
m_vec[ii] = np.mean(marginal_M[:, ii]) # get_mode
hpd_array_L[:, ii] = calcHPD(marginal_L[:, ii])
hpd_array_M[:, ii] = calcHPD(marginal_M[:, ii])
hpd_array_L50[:, ii] = calcHPD(marginal_L[:, ii], 0.75)
hpd_array_M50[:, ii] = calcHPD(marginal_M[:, ii], 0.75)
r_script += lib_utilities.print_R_vec("\n\nt", all_events)
r_script += "\ntime = -t"
r_script += lib_utilities.print_R_vec("\nspeciation", l_vec)
if i == -1:
r_script += lib_utilities.print_R_vec("\nsp_hdp_m", hpd_array_L[0])
r_script += lib_utilities.print_R_vec("\nsp_hdp_M", hpd_array_L[1])
r_script += lib_utilities.print_R_vec("\nsp_hdp_m50",
hpd_array_L50[0])
r_script += lib_utilities.print_R_vec("\nsp_hdp_M50",
hpd_array_L50[1])
r_script += lib_utilities.print_R_vec("\nextinction", m_vec)
l_vec= np.zeros(np.shape(marginal_L)[1])
m_vec= np.zeros(np.shape(marginal_L)[1])
hpd_array_L= np.zeros((2,np.shape(marginal_L)[1]))
hpd_array_M= np.zeros((2,np.shape(marginal_L)[1]))
hpd_array_L50= np.zeros((2,np.shape(marginal_L)[1]))
hpd_array_M50= np.zeros((2,np.shape(marginal_L)[1]))
if i>=0:
l_vec = np.mean(marginal_L, axis=0) # get_mode
m_vec = np.mean(marginal_M, axis=0) # get_mode
else:
for ii in range(np.shape(marginal_L)[1]): # loop over marginal rates
l_vec[ii] = np.mean(marginal_L[:,ii]) # get_mode
m_vec[ii] = np.mean(marginal_M[:,ii]) # get_mode
hpd_array_L[:,ii] = calcHPD(marginal_L[:,ii])
hpd_array_M[:,ii] = calcHPD(marginal_M[:,ii])
hpd_array_L50[:,ii] = calcHPD(marginal_L[:,ii],0.75)
hpd_array_M50[:,ii] = calcHPD(marginal_M[:,ii],0.75)
r_script = lib_utilities.print_R_vec("\n\nt",all_events)
r_script += "\ntime = -t"
r_script += lib_utilities.print_R_vec("\nspeciation",l_vec)
if i==-1:
r_script += lib_utilities.print_R_vec("\nsp_hdp_m",hpd_array_L[0])
r_script += lib_utilities.print_R_vec("\nsp_hdp_M",hpd_array_L[1])
r_script += lib_utilities.print_R_vec("\nsp_hdp_m50",hpd_array_L50[0])
r_script += lib_utilities.print_R_vec("\nsp_hdp_M50",hpd_array_L50[1])
r_script += lib_utilities.print_R_vec("\nextinction",m_vec)
if i==-1:
r_script += lib_utilities.print_R_vec("\nex_hdp_m",hpd_array_M[0])
Garray = np.array([Gl,Gm])*curve_scale_factor
age_vec,l_vec,m_vec = get_marginal_rates(args.m,L0,M0,Garray,Temp_at_events,shift_ind,root_age)
marginal_L.append(l_vec)
marginal_M.append(m_vec)
marginal_L = np.array(marginal_L)
marginal_M = np.array(marginal_M)
l_vec= np.zeros(np.shape(marginal_L)[1])
m_vec= np.zeros(np.shape(marginal_L)[1])
hpd_array_L= np.zeros((2,np.shape(marginal_L)[1]))
hpd_array_M= np.zeros((2,np.shape(marginal_L)[1]))
for i in range(np.shape(marginal_L)[1]):
l_vec[i] = np.mean(marginal_L[:,i])
m_vec[i] = np.mean(marginal_M[:,i])
hpd_array_L[:,i] = calcHPD(marginal_L[:,i])
hpd_array_M[:,i] = calcHPD(marginal_M[:,i])
print "done"
# write R file
print "\ngenerating R file...",
out="%s/%s_RTT.r" % (output_wd,name_file)
newfile = open(out, "wb")
if platform.system() == "Windows" or platform.system() == "Microsoft":
wd_forward = os.path.abspath(output_wd).replace('\\', '/')
r_script= "\n\npdf(file='%s/%s_RTT.pdf',width=0.6*20, height=0.6*20)\nlibrary(scales)\n" % (wd_forward,name_file)
else: r_script= "\n\npdf(file='%s/%s_RTT.pdf',width=0.6*20, height=0.6*20)\nlibrary(scales)\n" % (output_wd,name_file)
r_script += print_R_vec("\n\nt", age_vec)
r_script += "\ntime = -t"
r_script += print_R_vec("\nspeciation",l_vec)
r_script += print_R_vec("\nextinction",m_vec)
def RTTplot_high_res(f, grid_cell_size=1., burnin=0, max_age=0):
wd = "%s" % os.path.dirname(f)
name_file = os.path.splitext(os.path.basename(f))[0]
t = loadtxt(f, skiprows=max(1, int(burnin)))
head = np.array(next(open(f)).split()) # should be faster
#print np.where(head=="beta")[0], np.where(head=="temperature")[0]
if "temperature" in head or "beta" in head:
if "temperature" in head:
temp_index = np.where(head == "temperature")[0][0]
else:
temp_index = np.where(head == "beta")[0][0]
temp_values = t[:, temp_index]
t = t[temp_values == 1, :]
print("removed heated chains:", np.shape(t))
head = list(head)
sp_ind = [head.index(s) for s in head if "lambda_" in s]
ex_ind = [head.index(s) for s in head if "mu_" in s]
root_ind = head.index("root_age")
death_ind = head.index("death_age")
sp_shift_ind = [head.index(s) for s in head if "shift_sp_" in s]
ex_shift_ind = [head.index(s) for s in head if "shift_ex_" in s]
min_root_age = min(t[:, root_ind])
if max_age > 0: min_root_age = max_age
max_death_age = max(t[:, death_ind])
n_bins = int((min_root_age - max_death_age) / grid_cell_size)
grid = np.linspace(min_root_age, max_death_age, n_bins)
n_samples = np.shape(t)[0]
m_sp_matrix = np.zeros((n_samples, n_bins))
m_ex_matrix = np.zeros((n_samples, n_bins))
print("Extracting marginal rates...")
for i in range(n_samples):
l_shift_times = np.array([min_root_age] + list(t[i, sp_shift_ind]) +
[max_death_age])
l_rates = t[i, sp_ind]
m_sp_matrix[i] = get_marginal_rates_plot3(l_shift_times, l_rates, grid)
m_shift_times = np.array([min_root_age] + list(t[i, ex_shift_ind]) +
[max_death_age])
m_rates = t[i, ex_ind]
m_ex_matrix[i] = get_marginal_rates_plot3(m_shift_times, m_rates, grid)
res = np.zeros((n_bins, 7)) # times, l, lM, lm, m, mM, mm
print("Calculating HPDs...")
for i in range(n_bins):
l_HPD = list(util.calcHPD(m_sp_matrix[:, i], 0.95))
l_mean = mean(m_sp_matrix[:, i])
m_HPD = list(util.calcHPD(m_ex_matrix[:, i], 0.95))
m_mean = mean(m_ex_matrix[:, i])
res[i] = np.array([grid[i], l_mean] + l_HPD + [m_mean] + m_HPD)
Rfile = r_plot_code(res, wd, name_file)
out = "%s/%s_RTT.r" % (wd, name_file)
newfile = open(out, "w")
newfile.writelines(Rfile)
newfile.close()
print(
"\nAn R script with the source for the RTT plot was saved as: %sRTT.r\n(in %s)"
% (name_file, wd))
if platform.system() == "Windows" or platform.system() == "Microsoft":
cmd = "cd %s & Rscript %s_RTT.r" % (wd, name_file)
else:
cmd = "cd %s; Rscript %s/%s_RTT.r" % (wd, wd, name_file)
os.system(cmd)
save_HR_log_file = 1
if save_HR_log_file == 1:
print("Saving log file...")
out = "%s/%s_HR_marginal.log" % (wd, name_file)
logfile = open(out, "w")
head = ["iteration"] + ["l_%s" % (i) for i in range(n_bins)]
head += ["m_%s" % (i) for i in range(n_bins)]
wlog = csv.writer(logfile, delimiter='\t')
wlog.writerow(head)
for i in range(n_samples):
l = [i] + list(m_sp_matrix[i, :]) + list(m_ex_matrix[i, :])
wlog.writerow(l)
logfile.flush()
print("done\n")
def RTTplot_Q(f, q_shift_file, burnin=0, max_age=0):
wd = "%s" % os.path.dirname(f)
name_file = os.path.splitext(os.path.basename(f))[0]
t = loadtxt(f, skiprows=max(1, int(burnin)))
head = np.array(next(open(f)).split()) # should be faster
#print np.where(head=="beta")[0], np.where(head=="temperature")[0]
if "temperature" in head or "beta" in head:
if "temperature" in head:
temp_index = np.where(head == "temperature")[0][0]
else:
temp_index = np.where(head == "beta")[0][0]
temp_values = t[:, temp_index]
t = t[temp_values == 1, :]
print("removed heated chains:", np.shape(t))
head = list(head)
q_ind = [head.index(s) for s in head if "q_" in s]
root_ind = head.index("root_age")
death_ind = head.index("death_age")
min_root_age = min(t[:, root_ind])
if max_age > 0: min_root_age = max_age
max_death_age = max(t[:, death_ind])
try:
times_q_shift = np.sort(np.loadtxt(q_shift_file))[::-1]
except:
times_q_shift = np.array([np.loadtxt(q_shift_file)])
times_q_shift = times_q_shift[times_q_shift > max_death_age]
times_q_shift = times_q_shift[times_q_shift < min_root_age]
times_q_shift = np.sort(
np.array(list(times_q_shift) + [max_death_age, min_root_age]))[::-1]
print(times_q_shift)
means = []
hpdM = []
hpdm = []
data = "library(scales)\n"
if platform.system() == "Windows" or platform.system() == "Microsoft":
wd_forward = os.path.abspath(wd).replace('\\', '/')
data += "\n\npdf(file='%s/%s_RTT_Qrates.pdf',width=0.6*9, height=0.6*7)\n" % (
wd, name_file) # 9
else:
data += "\n\npdf(file='%s/%s_RTT_Qrates.pdf',width=0.6*9, height=0.6*7)\n" % (
wd, name_file) # 9
max_y_axis, max_x_axis, min_x_axis = np.max(
t[:, q_ind]), -np.max(times_q_shift), -np.min(times_q_shift)
for i in range(len(q_ind)):
qtemp = t[:, q_ind[i]]
hpdtemp = util.calcHPD(qtemp, 0.95)
#means.append(np.mean(qtemp),)
#hpdM.append(hpdtemp[1],hpdtemp[1])
#hpdm.append(hpdtemp[0],hpdtemp[0])
time_slice = np.array([times_q_shift[i], times_q_shift[i + 1]])
data += '\nage = c(%s, %s)' % (-time_slice[0], -time_slice[1])
data += '\nQ_mean = %s' % np.mean(qtemp)
data += '\nQ_hpd_m = %s' % hpdtemp[0]
data += '\nQ_hpd_M = %s' % hpdtemp[1]
if i == 0:
data += "\nplot(age,age,type = 'n', ylim = c(0, %s), xlim = c(%s,%s), ylab = 'Preservation rate', xlab = 'Ma',main='%s' )" \
% (max_y_axis,max_x_axis,min_x_axis,"Preservation rates")
else:
data += """\nsegments(x0=age[1], y0 = %s, x1 = age[1], y1 = Q_mean, col = "#756bb1", lwd=3)""" % (
Q_mean_previous)
Q_mean_previous = np.mean(qtemp)
data += """\nsegments(x0=age[1], y0 = Q_mean, x1 = age[2], y1 = Q_mean, col = "#756bb1", lwd=3)"""
data += """\npolygon( c(age, rev(age)), c(Q_hpd_m, Q_hpd_m, Q_hpd_M, Q_hpd_M), col = alpha("#756bb1",0.5), border = NA)"""
data += "\nn <- dev.off()"
out = "%s/%s_RTT_Qrates.r" % (wd, name_file)
newfile = open(out, "w")
newfile.writelines(data)
newfile.close()
print(
"\nAn R script with the source for the RTT plot was saved as: %s_RTT_Qrates.r\n(in %s)"
% (name_file, wd))
if platform.system() == "Windows" or platform.system() == "Microsoft":
cmd = "cd %s & Rscript %s_RTT_Qrates.r" % (wd, name_file)
else:
cmd = "cd %s; Rscript %s/%s_RTT_Qrates.r" % (wd, wd, name_file)
os.system(cmd)
print("done\n")
def RTTplot_high_res(f,grid_cell_size=1.,burnin=0,max_age=0):
wd = "%s" % os.path.dirname(f)
name_file=os.path.splitext(os.path.basename(f))[0]
t=loadtxt(f, skiprows=max(1,int(burnin)))
head = np.array(next(open(f)).split()) # should be faster
#print np.where(head=="beta")[0], np.where(head=="temperature")[0]
if "temperature" in head or "beta" in head:
if "temperature" in head:
temp_index = np.where(head=="temperature")[0][0]
else:
temp_index = np.where(head=="beta")[0][0]
temp_values = t[:,temp_index]
t = t[temp_values==1,:]
print "removed heated chains:",np.shape(t)
head= list(head)
sp_ind = [head.index(s) for s in head if "lambda_" in s]
ex_ind = [head.index(s) for s in head if "mu_" in s]
root_ind = head.index("root_age")
death_ind = head.index("death_age")
sp_shift_ind = [head.index(s) for s in head if "shift_sp_" in s]
ex_shift_ind = [head.index(s) for s in head if "shift_ex_" in s]
min_root_age = min(t[:,root_ind])
if max_age> 0: min_root_age=max_age
max_death_age= max(t[:,death_ind])
n_bins= int((min_root_age-max_death_age)/grid_cell_size)
grid = np.linspace(min_root_age,max_death_age,n_bins)
n_samples = np.shape(t)[0]
m_sp_matrix = np.zeros((n_samples,n_bins))
m_ex_matrix = np.zeros((n_samples,n_bins))
print "Extracting marginal rates..."
for i in range(n_samples):
l_shift_times = np.array([min_root_age]+ list(t[i,sp_shift_ind])+[max_death_age])
l_rates = t[i,sp_ind]
m_sp_matrix[i] = get_marginal_rates_plot3(l_shift_times,l_rates,grid)
m_shift_times = np.array([min_root_age]+ list(t[i,ex_shift_ind])+[max_death_age])
m_rates = t[i,ex_ind]
m_ex_matrix[i] = get_marginal_rates_plot3(m_shift_times,m_rates,grid)
res = np.zeros((n_bins,7)) # times, l, lM, lm, m, mM, mm
print "Calculating HPDs..."
for i in range(n_bins):
l_HPD = list(util.calcHPD(m_sp_matrix[:,i],0.95))
l_mean = mean(m_sp_matrix[:,i])
m_HPD = list(util.calcHPD(m_ex_matrix[:,i],0.95))
m_mean = mean(m_ex_matrix[:,i])
res[i] = np.array([grid[i],l_mean]+l_HPD+[m_mean]+m_HPD)
Rfile=r_plot_code(res,wd,name_file)
out="%s/%s_RTT.r" % (wd,name_file)
newfile = open(out, "w")
newfile.writelines(Rfile)
newfile.close()
print "\nAn R script with the source for the RTT plot was saved as: %sRTT.r\n(in %s)" % (name_file, wd)
if platform.system() == "Windows" or platform.system() == "Microsoft":
cmd="cd %s & Rscript %s_RTT.r" % (wd,name_file)
else:
cmd="cd %s; Rscript %s/%s_RTT.r" % (wd,wd,name_file)
os.system(cmd)
save_HR_log_file=1
if save_HR_log_file==1:
print "Saving log file..."
out="%s/%s_HR_marginal.log" % (wd,name_file)
logfile = open(out, "w")
head = ["iteration"]+["l_%s" % (i) for i in range(n_bins)]
head +=["m_%s" % (i) for i in range(n_bins)]
wlog=csv.writer(logfile, delimiter='\t')
wlog.writerow(head)
for i in range(n_samples):
l = [i]+ list(m_sp_matrix[i,:]) + list(m_ex_matrix[i,:])
wlog.writerow(l)
logfile.flush()
print "done\n"
def RTTplot_Q(f,q_shift_file,burnin=0,max_age=0):
wd = "%s" % os.path.dirname(f)
name_file=os.path.splitext(os.path.basename(f))[0]
t=loadtxt(f, skiprows=max(1,int(burnin)))
head = np.array(next(open(f)).split()) # should be faster
#print np.where(head=="beta")[0], np.where(head=="temperature")[0]
if "temperature" in head or "beta" in head:
if "temperature" in head:
temp_index = np.where(head=="temperature")[0][0]
else:
temp_index = np.where(head=="beta")[0][0]
temp_values = t[:,temp_index]
t = t[temp_values==1,:]
print "removed heated chains:",np.shape(t)
head= list(head)
q_ind = [head.index(s) for s in head if "q_" in s]
root_ind = head.index("root_age")
death_ind = head.index("death_age")
min_root_age = min(t[:,root_ind])
if max_age> 0: min_root_age=max_age
max_death_age= max(t[:,death_ind])
try: times_q_shift=np.sort(np.loadtxt(q_shift_file))[::-1]
except: times_q_shift=np.array([np.loadtxt(q_shift_file)])
times_q_shift = times_q_shift[times_q_shift>max_death_age]
times_q_shift = times_q_shift[times_q_shift<min_root_age]
times_q_shift = np.sort(np.array(list(times_q_shift) + [max_death_age,min_root_age]))[::-1]
print times_q_shift
means = []
hpdM = []
hpdm = []
data = "library(scales)\n"
if platform.system() == "Windows" or platform.system() == "Microsoft":
wd_forward = os.path.abspath(wd).replace('\\', '/')
data+= "\n\npdf(file='%s/%s_RTT_Qrates.pdf',width=0.6*9, height=0.6*7)\n" % (wd,name_file) # 9
else:
data+= "\n\npdf(file='%s/%s_RTT_Qrates.pdf',width=0.6*9, height=0.6*7)\n" % (wd,name_file) # 9
max_y_axis,max_x_axis,min_x_axis = np.max(t[:,q_ind]),-np.max(times_q_shift),-np.min(times_q_shift)
for i in range(len(q_ind)):
qtemp = t[:,q_ind[i]]
hpdtemp = util.calcHPD(qtemp,0.95)
#means.append(np.mean(qtemp),)
#hpdM.append(hpdtemp[1],hpdtemp[1])
#hpdm.append(hpdtemp[0],hpdtemp[0])
time_slice = np.array([times_q_shift[i],times_q_shift[i+1]])
data += '\nage = c(%s, %s)' % (-time_slice[0],-time_slice[1])
data += '\nQ_mean = %s' % np.mean(qtemp)
data += '\nQ_hpd_m = %s' % hpdtemp[0]
data += '\nQ_hpd_M = %s' % hpdtemp[1]
if i==0:
data += "\nplot(age,age,type = 'n', ylim = c(0, %s), xlim = c(%s,%s), ylab = 'Preservation rate', xlab = 'Ma',main='%s' )" \
% (max_y_axis,max_x_axis,min_x_axis,"Preservation rates")
else:
data += """\nsegments(x0=age[1], y0 = %s, x1 = age[1], y1 = Q_mean, col = "#756bb1", lwd=3)""" % (Q_mean_previous)
Q_mean_previous = np.mean(qtemp)
data += """\nsegments(x0=age[1], y0 = Q_mean, x1 = age[2], y1 = Q_mean, col = "#756bb1", lwd=3)"""
data += """\npolygon( c(age, rev(age)), c(Q_hpd_m, Q_hpd_m, Q_hpd_M, Q_hpd_M), col = alpha("#756bb1",0.5), border = NA)"""
data += "\nn <- dev.off()"
out="%s/%s_RTT_Qrates.r" % (wd,name_file)
newfile = open(out, "w")
newfile.writelines(data)
newfile.close()
print "\nAn R script with the source for the RTT plot was saved as: %s_RTT_Qrates.r\n(in %s)" % (name_file, wd)
if platform.system() == "Windows" or platform.system() == "Microsoft":
cmd="cd %s & Rscript %s_RTT_Qrates.r" % (wd,name_file)
else:
cmd="cd %s; Rscript %s/%s_RTT_Qrates.r" % (wd,wd,name_file)
os.system(cmd)
print "done\n"
