def ref_ali2d( ref_data ):
from utilities import print_msg
from filter import fit_tanh, filt_tanl
from utilities import center_2D
# Prepare the reference in 2D alignment, i.e., low-pass filter and center.
# Input: list ref_data
# 0 - mask
# 1 - center flag
# 2 - raw average
# 3 - fsc result
# Output: filtered, centered, and masked reference image
# apply filtration (FRC) to reference image:
global ref_ali2d_counter
ref_ali2d_counter += 1
print_msg("ref_ali2d #%6d\n"%(ref_ali2d_counter))
fl, aa = fit_tanh(ref_data[3])
aa = min(aa, 0.2)
fl = max(min(0.4,fl),0.12)
msg = "Tangent filter: cut-off frequency = %10.3f fall-off = %10.3f\n"%(fl, aa)
print_msg(msg)
tavg = filt_tanl(ref_data[2], fl, aa)
cs = [0.0]*2
if(ref_data[1] > 0):
tavg, cs[0], cs[1] = center_2D(tavg, ref_data[1], self_defined_reference = ref_data[0])
msg = "Center x = %10.3f Center y = %10.3f\n"%(cs[0], cs[1])
print_msg(msg)
return tavg, cs
def ref_7grp(ref_data):
from utilities import print_msg
from filter import fit_tanh, filt_tanl, filt_gaussinv
from fundamentals import fshift
from morphology import threshold
from math import sqrt
# Prepare the reference in 3D alignment, i.e., low-pass filter and center.
# Input: list ref_data
# 0 - mask
# 1 - center flag
# 2 - raw average
# 3 - fsc result
# Output: filtered, centered, and masked reference image
# apply filtration (FSC) to reference image:
#cs = [0.0]*3
stat = Util.infomask(ref_data[2], None, False)
volf = ref_data[2] - stat[0]
Util.mul_scalar(volf, 1.0 / stat[1])
volf = Util.muln_img(threshold(volf), ref_data[0])
fl, aa = fit_tanh(ref_data[3])
msg = "Tangent filter: cut-off frequency = %10.3f fall-off = %10.3f\n" % (
fl, aa)
print_msg(msg)
volf = filt_tanl(volf, fl, aa)
if (ref_data[1] == 1):
cs = volf.phase_cog()
msg = "Center x = %10.3f Center y = %10.3f Center z = %10.3f\n" % (
cs[0], cs[1], cs[2])
print_msg(msg)
volf = fshift(volf, -cs[0], -cs[1], -cs[2])
B_factor = 10.0
volf = filt_gaussinv(volf, 10.0)
return volf, cs
def ref_ali2d(ref_data):
from utilities import print_msg
from filter import fit_tanh, filt_tanl
from utilities import center_2D
# Prepare the reference in 2D alignment, i.e., low-pass filter and center.
# Input: list ref_data
# 0 - mask
# 1 - center flag
# 2 - raw average
# 3 - fsc result
# Output: filtered, centered, and masked reference image
# apply filtration (FRC) to reference image:
global ref_ali2d_counter
ref_ali2d_counter += 1
print_msg("ref_ali2d #%6d\n" % (ref_ali2d_counter))
fl, aa = fit_tanh(ref_data[3])
aa = min(aa, 0.2)
fl = max(min(0.4, fl), 0.12)
msg = "Tangent filter: cut-off frequency = %10.3f fall-off = %10.3f\n" % (
fl, aa)
print_msg(msg)
tavg = filt_tanl(ref_data[2], fl, aa)
cs = [0.0] * 2
if (ref_data[1] > 0):
tavg, cs[0], cs[1] = center_2D(tavg,
ref_data[1],
self_defined_reference=ref_data[0])
msg = "Center x = %10.3f Center y = %10.3f\n" % (
cs[0], cs[1])
print_msg(msg)
return tavg, cs
def ref_7grp( ref_data ): from utilities import print_msg from filter import fit_tanh, filt_tanl, filt_gaussinv from fundamentals import fshift from morphology import threshold from math import sqrt # Prepare the reference in 3D alignment, i.e., low-pass filter and center. # Input: list ref_data # 0 - mask # 1 - center flag # 2 - raw average # 3 - fsc result # Output: filtered, centered, and masked reference image # apply filtration (FSC) to reference image: #cs = [0.0]*3 stat = Util.infomask(ref_data[2], None, False) volf = ref_data[2] - stat[0] Util.mul_scalar(volf, 1.0/stat[1]) volf = Util.muln_img(threshold(volf), ref_data[0]) fl, aa = fit_tanh(ref_data[3]) msg = "Tangent filter: cut-off frequency = %10.3f fall-off = %10.3f\n"%(fl, aa) print_msg(msg) volf = filt_tanl(volf, fl, aa) if(ref_data[1] == 1): cs = volf.phase_cog() msg = "Center x = %10.3f Center y = %10.3f Center z = %10.3f\n"%(cs[0], cs[1], cs[2]) print_msg(msg) volf = fshift(volf, -cs[0], -cs[1], -cs[2]) B_factor = 10.0 volf = filt_gaussinv( volf, 10.0 ) return volf,cs
def filt_vols( vols, fscs, mask3D ): from math import sqrt from filter import fit_tanh, filt_tanl, filt_table from fundamentals import rops_table from morphology import threshold flmin = 1.0 flmax = -1.0 nvol = len(vols) for i in xrange(nvol): fl, aa = fit_tanh( fscs[i] ) if (fl < flmin): flmin = fl aamin = aa if (fl > flmax): flmax = fl idmax = i print " Filter tanl, parameters: ",flmin-0.05, " ", aamin volmax = vols[idmax] volmax = filt_tanl( volmax, flmin-0.05, aamin ) pmax = rops_table( volmax ) for i in xrange(nvol): ptab = rops_table( vols[i] ) for j in xrange( len(ptab) ): ptab[j] = sqrt( pmax[j]/ptab[j] ) vols[i] = filt_table( vols[i], ptab ) #stat = Util.infomask( vols[i], mask3D, False ) #volf -= stat[0] Util.mul_img( vols[i], mask3D ) #volf = threshold( volf ) return vols
def filt_vols( vols, fscs, mask3D ): from math import sqrt from filter import fit_tanh, filt_tanl, filt_table from fundamentals import rops_table from morphology import threshold flmin = 1.0 flmax = -1.0 nvol = len(vols) for i in xrange(nvol): fl, aa = fit_tanh( fscs[i] ) if (fl < flmin): flmin = fl aamin = aa if (fl > flmax): flmax = fl idmax = i print " Filter tanl, parameters: ",flmin-0.05, " ", aamin volmax = vols[idmax] volmax = filt_tanl( volmax, flmin-0.05, aamin ) pmax = rops_table( volmax ) for i in xrange(nvol): ptab = rops_table( vols[i] ) for j in xrange( len(ptab) ): ptab[j] = sqrt( pmax[j]/ptab[j] ) vols[i] = filt_table( vols[i], ptab ) #stat = Util.infomask( vols[i], mask3D, False ) #volf -= stat[0] Util.mul_img( vols[i], mask3D ) #volf = threshold( volf ) return vols
def ref_ali3d(ref_data):
from utilities import print_msg
from filter import fit_tanh, filt_tanl
from fundamentals import fshift
from morphology import threshold
# Prepare the reference in 3D alignment, i.e., low-pass filter and center.
# Input: list ref_data
# 0 - mask
# 1 - center flag
# 2 - raw average
# 3 - fsc result
# Output: filtered, centered, and masked reference image
# apply filtration (FSC) to reference image:
global ref_ali2d_counter
ref_ali2d_counter += 1
fl = ref_data[2].cmp("dot", ref_data[2], {
"negative": 0,
"mask": ref_data[0]
})
print_msg("ref_ali3d Step = %5d GOAL = %10.3e\n" %
(ref_ali2d_counter, fl))
cs = [0.0] * 3
#filt = filt_from_fsc(fscc, 0.05)
#vol = filt_table(vol, filt)
# here figure the filtration parameters and filter vol for the next iteration
#fl, fh = filt_params(res)
#vol = filt_btwl(vol, fl, fh)
# store the filtered reference volume
#lk = 0
#while(res[1][lk] >0.9 and res[0][lk]<0.25):
# lk+=1
#fl = res[0][lk]
#fh = min(fl+0.1,0.49)
#vol = filt_btwl(vol, fl, fh)
#fl, fh = filt_params(fscc)
#print "fl, fh, iter",fl,fh,Iter
#vol = filt_btwl(vol, fl, fh)
stat = Util.infomask(ref_data[2], ref_data[0], False)
volf = ref_data[2] - stat[0]
Util.mul_scalar(volf, 1.0 / stat[1])
#volf = threshold(volf)
Util.mul_img(volf, ref_data[0])
fl, aa = fit_tanh(ref_data[3])
#fl = 0.4
#aa = 0.1
msg = "Tangent filter: cut-off frequency = %10.3f fall-off = %10.3f\n" % (
fl, aa)
print_msg(msg)
volf = filt_tanl(volf, fl, aa)
if ref_data[1] == 1:
cs = volf.phase_cog()
msg = "Center x = %10.3f Center y = %10.3f Center z = %10.3f\n" % (
cs[0], cs[1], cs[2])
print_msg(msg)
volf = fshift(volf, -cs[0], -cs[1], -cs[2])
return volf, cs
def ref_ali3d( ref_data ):
from utilities import print_msg
from filter import fit_tanh, filt_tanl
from fundamentals import fshift
from morphology import threshold
# Prepare the reference in 3D alignment, i.e., low-pass filter and center.
# Input: list ref_data
# 0 - mask
# 1 - center flag
# 2 - raw average
# 3 - fsc result
# Output: filtered, centered, and masked reference image
# apply filtration (FSC) to reference image:
global ref_ali2d_counter
ref_ali2d_counter += 1
fl = ref_data[2].cmp("dot",ref_data[2], {"negative":0, "mask":ref_data[0]} )
print_msg("ref_ali3d Step = %5d GOAL = %10.3e\n"%(ref_ali2d_counter,fl))
cs = [0.0]*3
#filt = filt_from_fsc(fscc, 0.05)
#vol = filt_table(vol, filt)
# here figure the filtration parameters and filter vol for the next iteration
#fl, fh = filt_params(res)
#vol = filt_btwl(vol, fl, fh)
# store the filtered reference volume
#lk = 0
#while(res[1][lk] >0.9 and res[0][lk]<0.25):
# lk+=1
#fl = res[0][lk]
#fh = min(fl+0.1,0.49)
#vol = filt_btwl(vol, fl, fh)
#fl, fh = filt_params(fscc)
#print "fl, fh, iter",fl,fh,Iter
#vol = filt_btwl(vol, fl, fh)
stat = Util.infomask(ref_data[2], ref_data[0], False)
volf = ref_data[2] - stat[0]
Util.mul_scalar(volf, 1.0/stat[1])
#volf = threshold(volf)
Util.mul_img(volf, ref_data[0])
fl, aa = fit_tanh(ref_data[3])
#fl = 0.4
#aa = 0.1
msg = "Tangent filter: cut-off frequency = %10.3f fall-off = %10.3f\n"%(fl, aa)
print_msg(msg)
volf = filt_tanl(volf, fl, aa)
if ref_data[1] == 1:
cs = volf.phase_cog()
msg = "Center x = %10.3f Center y = %10.3f Center z = %10.3f\n"%(cs[0], cs[1], cs[2])
print_msg(msg)
volf = fshift(volf, -cs[0], -cs[1], -cs[2])
return volf, cs
def ref_aliB_cone(ref_data):
from utilities import print_msg
from filter import fit_tanh, filt_tanl
from fundamentals import fshift
from morphology import threshold
from math import sqrt
# Prepare the reference in 3D alignment, i.e., low-pass filter and center.
# Input: list ref_data
# 0 - mask
# 1 - reference PW
# 2 - raw average
# 3 - fsc result
# Output: filtered, centered, and masked reference image
# apply filtration (FSC) to reference image:
print_msg("ref_aliB_cone\n")
#cs = [0.0]*3
stat = Util.infomask(ref_data[2], None, True)
volf = ref_data[2] - stat[0]
Util.mul_scalar(volf, 1.0 / stat[1])
volf = threshold(volf)
Util.mul_img(volf, ref_data[0])
from fundamentals import rops_table
pwem = rops_table(volf)
ftb = []
for idum in xrange(len(pwem)):
ftb.append(sqrt(ref_data[1][idum] / pwem[idum]))
from filter import filt_table
volf = filt_table(volf, ftb)
fl, aa = fit_tanh(ref_data[3])
#fl = 0.41
#aa = 0.15
msg = "Tangent filter: cut-off frequency = %10.3f fall-off = %10.3f\n" % (
fl, aa)
print_msg(msg)
volf = filt_tanl(volf, fl, aa)
stat = Util.infomask(volf, None, True)
volf -= stat[0]
Util.mul_scalar(volf, 1.0 / stat[1])
"""
if(ref_data[1] == 1):
cs = volf.phase_cog()
msg = "Center x = %10.3f Center y = %10.3f Center z = %10.3f\n"%(cs[0], cs[1], cs[2])
print_msg(msg)
volf = fshift(volf, -cs[0], -cs[1], -cs[2])
"""
return volf
def ref_aliB_cone( ref_data ):
from utilities import print_msg
from filter import fit_tanh, filt_tanl
from fundamentals import fshift
from morphology import threshold
from math import sqrt
# Prepare the reference in 3D alignment, i.e., low-pass filter and center.
# Input: list ref_data
# 0 - mask
# 1 - reference PW
# 2 - raw average
# 3 - fsc result
# Output: filtered, centered, and masked reference image
# apply filtration (FSC) to reference image:
print_msg("ref_aliB_cone\n")
#cs = [0.0]*3
stat = Util.infomask(ref_data[2], None, True)
volf = ref_data[2] - stat[0]
Util.mul_scalar(volf, 1.0/stat[1])
volf = threshold(volf)
Util.mul_img(volf, ref_data[0])
from fundamentals import rops_table
pwem = rops_table(volf)
ftb = []
for idum in xrange(len(pwem)):
ftb.append(sqrt(ref_data[1][idum]/pwem[idum]))
from filter import filt_table
volf = filt_table(volf, ftb)
fl, aa = fit_tanh(ref_data[3])
#fl = 0.41
#aa = 0.15
msg = "Tangent filter: cut-off frequency = %10.3f fall-off = %10.3f\n"%(fl, aa)
print_msg(msg)
volf = filt_tanl(volf, fl, aa)
stat = Util.infomask(volf, None, True)
volf -= stat[0]
Util.mul_scalar(volf, 1.0/stat[1])
"""
if(ref_data[1] == 1):
cs = volf.phase_cog()
msg = "Center x = %10.3f Center y = %10.3f Center z = %10.3f\n"%(cs[0], cs[1], cs[2])
print_msg(msg)
volf = fshift(volf, -cs[0], -cs[1], -cs[2])
"""
return volf
def minfilt( fscc ): from filter import fit_tanh numref = len(fscc) flmin = 1.0 flmax = -1.0 for iref in xrange(numref): fl, aa = fit_tanh( fscc[iref] ) if (fl < flmin): flmin = fl aamin = aa idmin = iref if (fl > flmax): flmax = fl aamax = aa return flmin,aamin,idmin
def minfilt(fscc):
from filter import fit_tanh
numref = len(fscc)
flmin = 1.0
flmax = -1.0
for iref in xrange(numref):
fl, aa = fit_tanh(fscc[iref])
if (fl < flmin):
flmin = fl
aamin = aa
idmin = iref
if (fl > flmax):
flmax = fl
aamax = aa
return flmin, aamin, idmin
def ref_ali3dm_ali_50S(refdata):
from filter import fit_tanh, filt_tanl
from utilities import get_im
from fundamentals import rot_shift3D
import os
numref = refdata[0]
outdir = refdata[1]
fscc = refdata[2]
total_iter = refdata[3]
varf = refdata[4]
#mask_50S = get_im( "mask-50S.spi" )
flmin = 1.0
flmax = -1.0
for iref in xrange(numref):
fl, aa = fit_tanh(fscc[iref])
if (fl < flmin):
flmin = fl
aamin = aa
if (fl > flmax):
flmax = fl
aamax = aa
print 'iref,fl,aa: ', iref, fl, aa
# filter to minimum resolution
print 'flmin,aamin:', flmin, aamin
for iref in xrange(numref):
v = get_im(os.path.join(outdir, "vol%04d.hdf" % total_iter), iref)
v = filt_tanl(v, flmin, aamin)
if ali50s:
from utilities import get_params3D, set_params3D, combine_params3
from applications import ali_vol_shift, ali_vol_rotate
if iref == 0:
v50S_ref = alivol_mask_getref(v, mask_50S)
else:
v = alivol_mask(v, v50S_ref, mask_50S)
if not (varf is None):
print 'filtering by fourier variance'
v.filter_by_image(varf)
v.write_image(os.path.join(outdir, "volf%04d.hdf" % total_iter), iref)
def ref_ali3dm_ali_50S( refdata ): from filter import fit_tanh, filt_tanl from utilities import get_im from fundamentals import rot_shift3D import os numref = refdata[0] outdir = refdata[1] fscc = refdata[2] total_iter = refdata[3] varf = refdata[4] #mask_50S = get_im( "mask-50S.spi" ) flmin = 1.0 flmax = -1.0 for iref in xrange(numref): fl, aa = fit_tanh( fscc[iref] ) if (fl < flmin): flmin = fl aamin = aa if (fl > flmax): flmax = fl aamax = aa print 'iref,fl,aa: ', iref, fl, aa # filter to minimum resolution print 'flmin,aamin:', flmin, aamin for iref in xrange(numref): v = get_im(os.path.join(outdir, "vol%04d.hdf"%total_iter), iref) v = filt_tanl(v, flmin, aamin) if ali50s: from utilities import get_params3D, set_params3D, combine_params3 from applications import ali_vol_shift, ali_vol_rotate if iref==0: v50S_ref = alivol_mask_getref( v, mask_50S ) else: v = alivol_mask( v, v50S_ref, mask_50S ) if not(varf is None): print 'filtering by fourier variance' v.filter_by_image( varf ) v.write_image(os.path.join(outdir, "volf%04d.hdf"%total_iter), iref)
def spruce_up(ref_data):
from utilities import print_msg
from filter import filt_tanl, fit_tanh
from morphology import threshold
# Prepare the reference in 3D alignment, i.e., low-pass filter and center.
# Input: list ref_data
# 0 - mask
# 1 - center flag
# 2 - raw average
# 3 - fsc result
# Output: filtered, centered, and masked reference image
# apply filtration (FSC) to reference image:
print_msg("Changed4 spruce_up\n")
cs = [0.0] * 3
stat = Util.infomask(ref_data[2], None, True)
volf = ref_data[2] - stat[0]
Util.mul_scalar(volf, 1.0 / stat[1])
volf = threshold(volf)
# Apply B-factor
from filter import filt_gaussinv
from math import sqrt
B = 1.0 / sqrt(2. * 14.0)
volf = filt_gaussinv(volf, B, False)
nx = volf.get_xsize()
from utilities import model_circle
stat = Util.infomask(
volf,
model_circle(nx // 2 - 2, nx, nx, nx) -
model_circle(nx // 2 - 6, nx, nx, nx), True)
volf -= stat[0]
Util.mul_img(volf, ref_data[0])
fl, aa = fit_tanh(ref_data[3])
#fl = 0.35
#aa = 0.1
aa /= 2
msg = "Tangent filter: cut-off frequency = %10.3f fall-off = %10.3f\n" % (
fl, aa)
print_msg(msg)
volf = filt_tanl(volf, fl, aa)
return volf, cs
def ref_ali3d( ref_data ):
from utilities import print_msg
from filter import fit_tanh, filt_tanl
from fundamentals import fshift
from morphology import threshold
fl = ref_data[2].cmp("dot",ref_data[2], {"negative":0, "mask":ref_data[0]} )
cs = [0.0]*3
stat = Util.infomask(ref_data[2], ref_data[0], False)
volf = ref_data[2] - stat[0]
Util.mul_scalar(volf, 1.0/stat[1])
Util.mul_img(volf, ref_data[0])
fl, aa = fit_tanh(ref_data[3])
volf = filt_tanl(volf, fl, aa)
volf.process_inplace("normalize")
if ref_data[1] == 1:
cs = volf.phase_cog()
volf = fshift(volf, -cs[0], -cs[1], -cs[2])
return volf, cs, fl
def spruce_up( ref_data ):
from utilities import print_msg
from filter import filt_tanl, fit_tanh
from morphology import threshold
# Prepare the reference in 3D alignment, i.e., low-pass filter and center.
# Input: list ref_data
# 0 - mask
# 1 - center flag
# 2 - raw average
# 3 - fsc result
# Output: filtered, centered, and masked reference image
# apply filtration (FSC) to reference image:
print_msg("Changed4 spruce_up\n")
cs = [0.0]*3
stat = Util.infomask(ref_data[2], None, True)
volf = ref_data[2] - stat[0]
Util.mul_scalar(volf, 1.0/stat[1])
volf = threshold(volf)
# Apply B-factor
from filter import filt_gaussinv
from math import sqrt
B = 1.0/sqrt(2.*14.0)
volf = filt_gaussinv(volf, B, False)
nx = volf.get_xsize()
from utilities import model_circle
stat = Util.infomask(volf, model_circle(nx//2-2,nx,nx,nx)-model_circle(nx//2-6,nx,nx,nx), True)
volf -= stat[0]
Util.mul_img(volf, ref_data[0])
fl, aa = fit_tanh(ref_data[3])
#fl = 0.35
#aa = 0.1
aa /= 2
msg = "Tangent filter: cut-off frequency = %10.3f fall-off = %10.3f\n"%(fl, aa)
print_msg(msg)
volf = filt_tanl(volf, fl, aa)
return volf, cs
def ref_ali3d(ref_data):
from utilities import print_msg
from filter import fit_tanh, filt_tanl
from fundamentals import fshift
from morphology import threshold
fl = ref_data[2].cmp("dot", ref_data[2], {
"negative": 0,
"mask": ref_data[0]
})
cs = [0.0] * 3
stat = Util.infomask(ref_data[2], ref_data[0], False)
volf = ref_data[2] - stat[0]
Util.mul_scalar(volf, 1.0 / stat[1])
Util.mul_img(volf, ref_data[0])
fl, aa = fit_tanh(ref_data[3])
volf = filt_tanl(volf, fl, aa)
volf.process_inplace("normalize")
if ref_data[1] == 1:
cs = volf.phase_cog()
volf = fshift(volf, -cs[0], -cs[1], -cs[2])
return volf, cs, fl
def compare(compare_ref_free, outfile_repro, ref_free_output, yrng, xrng,
rstep, nx, apix, ref_free_cutoff, nproc, myid, main_node):
from alignment import Numrinit, ringwe, Applyws
from random import seed, randint
from utilities import get_params2D, set_params2D, model_circle, inverse_transform2, combine_params2
from fundamentals import rot_shift2D
from mpi import MPI_COMM_WORLD, mpi_barrier, mpi_bcast, MPI_INT
from statistics import fsc_mask
from filter import fit_tanh
from numpy import array
fout = "%s.hdf" % ref_free_output
frc_out = "%s_frc" % ref_free_output
res_out = "%s_res" % ref_free_output
nima = EMUtil.get_image_count(compare_ref_free)
image_start, image_end = MPI_start_end(nima, nproc, myid)
ima = EMData()
ima.read_image(compare_ref_free, image_start)
last_ring = nx / 2 - 2
first_ring = 1
mask = model_circle(last_ring, nx, nx)
refi = []
numref = EMUtil.get_image_count(outfile_repro)
cnx = nx / 2 + 1
cny = cnx
mode = "F"
numr = Numrinit(first_ring, last_ring, rstep, mode)
wr = ringwe(numr, mode)
ima.to_zero()
for j in xrange(numref):
temp = EMData()
temp.read_image(outfile_repro, j)
# even, odd, numer of even, number of images. After frc, totav
refi.append(temp)
# for each node read its share of data
data = EMData.read_images(compare_ref_free, range(image_start, image_end))
for im in xrange(image_start, image_end):
data[im - image_start].set_attr('ID', im)
set_params2D(data[im - image_start], [0, 0, 0, 0, 1])
ringref = []
for j in xrange(numref):
refi[j].process_inplace("normalize.mask", {
"mask": mask,
"no_sigma": 1
}) # normalize reference images to N(0,1)
cimage = Util.Polar2Dm(refi[j], cnx, cny, numr, mode)
Util.Frngs(cimage, numr)
Applyws(cimage, numr, wr)
ringref.append(cimage)
if myid == main_node: seed(1000)
data_shift = []
frc = []
res = []
for im in xrange(image_start, image_end):
alpha, sx, sy, mirror, scale = get_params2D(data[im - image_start])
alphai, sxi, syi, scalei = inverse_transform2(alpha, sx, sy, 1.0)
# normalize
data[im - image_start].process_inplace("normalize.mask", {
"mask": mask,
"no_sigma": 1
}) # subtract average under the mask
# align current image to the reference
[angt, sxst, syst, mirrort, xiref,
peakt] = Util.multiref_polar_ali_2d(data[im - image_start], ringref,
xrng, yrng, 1, mode, numr,
cnx + sxi, cny + syi)
iref = int(xiref)
[alphan, sxn, syn, mn] = combine_params2(0.0, -sxi, -syi, 0, angt,
sxst, syst, (int)(mirrort))
set_params2D(data[im - image_start],
[alphan, sxn, syn, int(mn), scale])
temp = rot_shift2D(data[im - image_start], alphan, sxn, syn, mn)
temp.set_attr('assign', iref)
tfrc = fsc_mask(temp, refi[iref], mask=mask)
temp.set_attr('frc', tfrc[1])
res = fit_tanh(tfrc)
temp.set_attr('res', res)
data_shift.append(temp)
for node in xrange(nproc):
if myid == node:
for image in data_shift:
image.write_image(fout, -1)
refindex = image.get_attr('assign')
refi[refindex].write_image(fout, -1)
mpi_barrier(MPI_COMM_WORLD)
rejects = []
if myid == main_node:
a = EMData()
index = 0
frc = []
res = []
temp = []
classes = []
for im in xrange(nima):
a.read_image(fout, index)
frc.append(a.get_attr("frc"))
if ref_free_cutoff != -1:
classes.append(a.get_attr("class_ptcl_idxs"))
tmp = a.get_attr("res")
temp.append(tmp[0])
res.append("%12f" % (apix / tmp[0]))
res.append("\n")
index = index + 2
res_num = array(temp)
mean_score = res_num.mean(axis=0)
std_score = res_num.std(axis=0)
std = std_score / 2
if ref_free_cutoff != -1:
cutoff = mean_score - std * ref_free_cutoff
reject = res_num < cutoff
index = 0
for i in reject:
if i: rejects.extend(classes[index])
index = index + 1
rejects.sort()
length = mpi_bcast(len(rejects), 1, MPI_INT, main_node,
MPI_COMM_WORLD)
rejects = mpi_bcast(rejects, length, MPI_INT, main_node,
MPI_COMM_WORLD)
del a
fout_frc = open(frc_out, 'w')
fout_res = open(res_out, 'w')
fout_res.write("".join(res))
temp = zip(*frc)
datstrings = []
for i in temp:
for j in i:
datstrings.append(" %12f" % (j))
datstrings.append("\n")
fout_frc.write("".join(datstrings))
fout_frc.close()
del refi
del ringref
return rejects
def compare(compare_ref_free, outfile_repro,ref_free_output,yrng, xrng, rstep,nx,apix,ref_free_cutoff, nproc, myid, main_node):
from alignment import Numrinit, ringwe, Applyws
from random import seed, randint
from utilities import get_params2D, set_params2D, model_circle, inverse_transform2, combine_params2
from fundamentals import rot_shift2D
from mpi import MPI_COMM_WORLD, mpi_barrier, mpi_bcast, MPI_INT
from statistics import fsc_mask
from filter import fit_tanh
from numpy import array
fout = "%s.hdf" % ref_free_output
frc_out = "%s_frc" % ref_free_output
res_out = "%s_res" % ref_free_output
nima = EMUtil.get_image_count(compare_ref_free)
image_start, image_end = MPI_start_end(nima, nproc, myid)
ima = EMData()
ima.read_image(compare_ref_free, image_start)
last_ring = nx/2-2
first_ring = 1
mask = model_circle(last_ring, nx, nx)
refi = []
numref = EMUtil.get_image_count(outfile_repro)
cnx = nx/2 +1
cny = cnx
mode = "F"
numr = Numrinit(first_ring, last_ring, rstep, mode)
wr = ringwe(numr, mode)
ima.to_zero()
for j in xrange(numref):
temp = EMData()
temp.read_image(outfile_repro, j)
# even, odd, numer of even, number of images. After frc, totav
refi.append(temp)
# for each node read its share of data
data = EMData.read_images(compare_ref_free, range(image_start, image_end))
for im in xrange(image_start, image_end):
data[im-image_start].set_attr('ID', im)
set_params2D(data[im-image_start],[0,0,0,0,1])
ringref = []
for j in xrange(numref):
refi[j].process_inplace("normalize.mask", {"mask":mask, "no_sigma":1}) # normalize reference images to N(0,1)
cimage = Util.Polar2Dm(refi[j], cnx, cny, numr, mode)
Util.Frngs(cimage, numr)
Applyws(cimage, numr, wr)
ringref.append(cimage)
if myid == main_node: seed(1000)
data_shift = []
frc = []
res = []
for im in xrange(image_start, image_end):
alpha, sx, sy, mirror, scale = get_params2D(data[im-image_start])
alphai, sxi, syi, scalei = inverse_transform2(alpha, sx, sy, 1.0)
# normalize
data[im-image_start].process_inplace("normalize.mask", {"mask":mask, "no_sigma":1}) # subtract average under the mask
# align current image to the reference
[angt, sxst, syst, mirrort, xiref, peakt] = Util.multiref_polar_ali_2d(data[im-image_start], ringref, xrng, yrng, 1, mode, numr, cnx+sxi, cny+syi)
iref = int(xiref)
[alphan, sxn, syn, mn] = combine_params2(0.0, -sxi, -syi, 0, angt, sxst, syst, (int)(mirrort))
set_params2D(data[im-image_start], [alphan, sxn, syn, int(mn), scale])
temp = rot_shift2D(data[im-image_start], alphan, sxn, syn, mn)
temp.set_attr('assign',iref)
tfrc = fsc_mask(temp,refi[iref],mask = mask)
temp.set_attr('frc',tfrc[1])
res = fit_tanh(tfrc)
temp.set_attr('res',res)
data_shift.append(temp)
for node in xrange(nproc):
if myid == node:
for image in data_shift:
image.write_image(fout,-1)
refindex = image.get_attr('assign')
refi[refindex].write_image(fout,-1)
mpi_barrier(MPI_COMM_WORLD)
rejects = []
if myid == main_node:
a = EMData()
index = 0
frc = []
res = []
temp = []
classes = []
for im in xrange(nima):
a.read_image(fout, index)
frc.append(a.get_attr("frc"))
if ref_free_cutoff != -1: classes.append(a.get_attr("class_ptcl_idxs"))
tmp = a.get_attr("res")
temp.append(tmp[0])
res.append("%12f" %(apix/tmp[0]))
res.append("\n")
index = index + 2
res_num = array(temp)
mean_score = res_num.mean(axis=0)
std_score = res_num.std(axis=0)
std = std_score / 2
if ref_free_cutoff !=-1:
cutoff = mean_score - std * ref_free_cutoff
reject = res_num < cutoff
index = 0
for i in reject:
if i: rejects.extend(classes[index])
index = index + 1
rejects.sort()
length = mpi_bcast(len(rejects),1,MPI_INT,main_node, MPI_COMM_WORLD)
rejects = mpi_bcast(rejects,length , MPI_INT, main_node, MPI_COMM_WORLD)
del a
fout_frc = open(frc_out,'w')
fout_res = open(res_out,'w')
fout_res.write("".join(res))
temp = zip(*frc)
datstrings = []
for i in temp:
for j in i:
datstrings.append(" %12f" % (j))
datstrings.append("\n")
fout_frc.write("".join(datstrings))
fout_frc.close()
del refi
del ringref
return rejects