def plot_height_profiles():
"""
Plot the bilayer height profile with protein positions.
"""
#---one plot per simulation
for sn in work.sns():
mesh = data[sn]['data']['mesh']
surf = mesh.mean(axis=0).mean(axis=0)
surf -= surf.mean()
hmax = np.abs(surf).max()
fig = plt.figure()
ax = fig.add_subplot(111)
vecs = data[sn]['data']['vecs'].mean(axis=0)
im = ax.imshow(surf.T,origin='lower',
interpolation='nearest',cmap=mpl.cm.__dict__['RdBu_r'],
extent=[0,vecs[0],0,vecs[1]],vmax=hmax,vmin=-1*hmax)
#---! add a check for whether we have proteins or not
if sn in data_prot:
from render.wavevids import plothull
points_all = data_prot[sn]['data']['points_all']
points_all_mean_time = points_all.mean(axis=0)
plothull(ax,points_all_mean_time[...,:2],griddims=surf.shape,vecs=vecs,c='k',lw=0)
from mpl_toolkits.axes_grid1.inset_locator import inset_axes
axins = inset_axes(ax,width="5%",height="100%",loc=3,
bbox_to_anchor=(1.05,0.,1.,1.),bbox_transform=ax.transAxes,borderpad=0)
cbar = plt.colorbar(im,cax=axins,orientation="vertical")
cbar.set_label(r"$\mathrm{\langle z \rangle (nm)}$",labelpad=-40,y=1.05,rotation=0)
axins.tick_params(axis='y',which='both',left='off',right='off',labelright='on')
ax.set_title('average bilayer height')
ax.set_xlabel('x (nm)')
ax.set_ylabel('y (nm)')
ax.tick_params(axis='y',which='both',left='off',right='off',labelleft='on')
ax.tick_params(axis='x',which='both',top='off',bottom='off',labelbottom='on')
picturesave('fig.height_average.%s'%sn,work.plotdir,backup=False,version=True,meta={})
def plot_hull_and_trial_centers(data,
sn,
ax,
n_instances=None,
debug_frame=0,
color=None):
"""Plot the protein hull along with the positions of the trial functions."""
global postdat
#---! currently only set for a single protein
trials = data[sn]['drop_gaussians_points'].transpose(1, 0, 2)
nframes = len(trials)
if n_instances != None:
samples = np.arange(0, nframes, nframes / n_instances)
else:
samples = np.array([debug_frame])
pts = np.concatenate(trials[samples])
for ptsnum, pts in enumerate(trials[samples]):
color_this = mpl.cm.__dict__['jet'](
float(ptsnum) / len(samples)) if not color else color
ax.scatter(*pts.T, s=1, c=color_this)
griddims = data[sn]['cf'].shape
vecs, points_protein = [postdat[sn][i] for i in ['vecs', 'points_protein']]
for ptsnum, pts in enumerate(points_protein[samples][..., :2]):
color_this = mpl.cm.__dict__['jet'](
float(ptsnum) / len(samples)) if not color else color
plothull(ax, pts, griddims=griddims, vecs=vecs, lw=0, c=color_this)
ax.set_xlim((0, vecs[0]))
ax.set_ylim((0, vecs[1]))
ax.set_aspect(1.0)
def plothull_protein_center_average(ax, sn, vecs, tag):
"""Plot the protein hull in the center of the box."""
dat_protein = data[protein_abstractor_name][sn]['data']
protein_noh = np.array([
ii for ii, i in enumerate(dat_protein['names'])
if not re.match('^H', i)
])
protein_pts = dat_protein['points_all']
if protein_pts.shape[1] != 1:
raise Exception('cannot center multiple proteins!')
mean_prot_pts = protein_pts[:, 0].mean(axis=0)[:, :2]
prot_center = mean_prot_pts - mean_prot_pts.mean(axis=0) + vecs[:2] / 2.
plothull(ax, [prot_center],
griddims=datas[tag][sn]['cf'].shape,
vecs=vecs,
c='k',
lw=0)
def render_hydrogen_bonding_pattern(fr, frameno):
"""..."""
#status('drawing for %s'%sn,i=fr,looplen=nframes,tag='render')
#---copy points so that we can smooth the points in the box
points = np.array(lipid_mesh['%d.%d.points' % (top_mono, fr)])
if smooth_window: points[:nmol] = points_inside_smooth[fr]
ax = plt.subplot(111)
vec = dat_lipids['vecs'][fr]
#---get the hydrogen bonds for this plot
resname_inds = np.array([
bond_spec_keys.split().index(j)
for j in ['resid_%d' % i for i in [1, 2]]
])
#---use frameno here because obs is over valid_frames only
bonds_this_all = bonds[np.where(obs[frameno])[0]][:,
resname_inds].astype(int)
#---filter out intramolecular hyrogen bonds
bonds_this = bonds_this_all[np.where(
bonds_this_all[:, 0] != bonds_this_all[:, 1])[0]]
#---! note no check for resname type
#---get bonds that are in the mesh
bonds_this_valid = np.where(
[np.all([np.in1d(i, r2m.keys()) for i in b]) for b in bonds_this])[0]
ghosts = lipid_mesh['%d.%d.ghost_ids' % (top_mono, fr)]
#---draw a line for each bond
for lnum, link_resids in enumerate(bonds_this[bonds_this_valid]):
verts = np.array([r2m[j] for j in link_resids])
points_this = points[verts, :2]
if not np.any(points_this.ptp(axis=0) >= vec[:2] / 2.):
ax.plot(*points_this.T, c='k', lw=1, zorder=2)
else:
targs = np.where(np.in1d(ghosts, verts))[0]
for target in targs:
for urp in link_resids:
possible_long_bond = np.array([target, r2m[urp]])
try:
if not np.any((points[possible_long_bond].ptp(axis=0)
>= vec / 2.)[:2]):
ax.plot(*points[possible_long_bond][:, :2].T,
c='k',
lw=1,
zorder=2)
except:
pass
#---formulate the color list
colors = np.array(
[colorize(work.meta[sn], resname=r) for r in resnames_all[ghosts]])
#---stray dots are transparent
ax.scatter(*points[:, :2].T,
s=dot_size_small,
color=colors,
alpha=0.35,
zorder=3,
lw=0)
#---active dots have color
highlit = np.array(
[r2m[j] for j in np.unique(bonds_this[bonds_this_valid])])
#---note that only the non-ghost points can be highlit so the color list below is complete
if highlit.max() >= m2r.shape[0]:
raise Exception('some highlit points are not in the mesh keys!')
ax.scatter(
*points[highlit, :2].T,
lw=0.25,
s=dot_size_small,
zorder=4,
edgecolor='k',
c=[colorize(work.meta[sn], resname=r) for r in resnames_all[highlit]])
#---highlight the PIP2
highlit_pip2 = np.where(
np.in1d(resnames_all[highlit],
work.vars['selectors']['resnames_PIP2'] + ['PtdIns']))[0]
ax.scatter(*points[highlit[highlit_pip2], :2].T,
lw=0.25,
s=dot_size_large,
zorder=5,
edgecolor='k',
c=[
colorize(work.meta[sn], resname=r)
for r in resnames_all[highlit[highlit_pip2]]
])
#---show protein hull if necessary
if show_proteins:
plothull(ax, [data_prot[sn]['data']['points'][fr][:, :2]],
vecs=data_prot[sn]['data']['vecs'][fr],
griddims=None,
fill=True,
lw=0,
alpha=0.35,
c='k')
#---plot a box
lims_prop = 0.1
ax.set_xlim((vec[0] * -lims_prop, vec[0] * (1. + lims_prop)))
ax.set_ylim((vec[1] * -lims_prop, vec[1] * (1. + lims_prop)))
box_prop = (lims_prop * 2 + 1.0) / lims_prop
ax.axhline(0,
xmin=1. / box_prop,
xmax=(box_prop - 1.0) / box_prop,
c='k',
lw=1)
ax.axhline(vec[1],
xmin=1. / box_prop,
xmax=(box_prop - 1.0) / (box_prop),
c='k',
lw=1)
ax.axvline(0,
ymin=1. / box_prop,
ymax=(box_prop - 1.0) / (box_prop),
c='k',
lw=1)
ax.axvline(vec[0],
ymin=1. / box_prop,
ymax=(box_prop - 1.0) / (box_prop),
c='k',
lw=1)
ax.set_aspect('equal')
ax.set_xticks([])
ax.set_yticks([])
ax.tick_params(axis=u'both', which=u'both', length=0)
ax.axis('off')
picturesave('snap.%05d' % frameno,
out_dn,
backup=False,
version=True,
meta={},
extras=[],
loud=False)
def plot_curvature(sn, **kwargs):
"""
Plot a map of the curvature.
"""
global curvs_map_all
dat = data[sn]['data']
nframes = int(dat['nframes'])
def get(mn, fr, name):
return dat['%d.%d.%s' % (mn, fr, name)]
spacing = kwargs.pop('spacing', 1.0)
if kwargs: raise Exception('unprocessed kwargs %s' % kwargs)
mvecs = np.mean([get(0, fr, 'vec') for fr in range(nframes)], axis=0)
ngrid = np.round(mvecs / spacing).astype(int)
#---curvature of each leaflet
curvs_map_all = [np.zeros((ngrid[0], ngrid[1])) for mn in range(2)]
for mn in range(2):
curvs = np.zeros((ngrid[0], ngrid[1]))
curvs_counts = np.zeros((ngrid[0], ngrid[1]))
nmol = int(dat['%d.1.nmol' % mn])
for fr in range(nframes):
simps = get(mn, fr, 'simplices')
pts = get(mn, fr, 'points')
vec = get(mn, fr, 'vec')
curv = get(mn, fr, 'mean')
#---repack the points in the box
pts_repack = (pts - np.floor(pts / vec) * vec)
pts_rounded = (pts_repack / (vec / ngrid)).astype(int)[:, :2]
curvs[pts_rounded[:nmol, 0], pts_rounded[:nmol, 1]] += curv[:nmol]
curvs_counts[pts_rounded[:nmol, 0], pts_rounded[:nmol, 1]] += 1
obs = np.where(curvs_counts > 0)
means = curvs[obs] / curvs_counts[obs]
curvs_map_all[mn][obs] = means
from mpl_toolkits.axes_grid1.inset_locator import inset_axes
import scipy
import scipy.ndimage
#---average the leaflets
curvs_map = np.mean(curvs_map_all, axis=0)
#---! removed the following line which sets nan (white on imshow). not sure what it is for
#---ironically back to nan for the imshow
if False: curvs_map[np.where(curvs_counts == 0)] = np.nan
#---! it might be nice to smooth the plots but beware!
if False:
curvs_map = scipy.ndimage.gaussian_filter(curvs_map,
sigma=(4, 4),
order=0)
vmax = max([abs(j) for j in [curvs_map.max(), curvs_map.min()]])
ax = plt.gca()
im = ax.imshow(curvs_map,
cmap=mpl.cm.RdBu_r,
vmax=vmax,
vmin=-1 * vmax,
origin='lower',
extent=[0, mvecs[0], 0, mvecs[1]])
axins = inset_axes(ax,
width="5%",
height="100%",
loc=3,
bbox_to_anchor=(1.05, 0., 1., 1.),
bbox_transform=ax.transAxes,
borderpad=0)
cbar = plt.colorbar(im, cax=axins, orientation="vertical")
cbar.set_label(r'$C_0\,({nm}^{-1})$', labelpad=-40, y=1.05, rotation=0)
ax.set_xlabel('x (nm)')
ax.set_ylabel('y (nm)')
ax.tick_params(axis='y',
which='both',
left='off',
right='off',
labelleft='on')
ax.tick_params(axis='x',
which='both',
top='off',
bottom='off',
labelbottom='on')
axins.tick_params(axis='y',
which='both',
left='off',
right='off',
labelright='on')
ax.set_title('curvature map')
if sn in data_prot:
from render.wavevids import plothull
points_all = data_prot[sn]['data']['points_all']
points_all_mean_time = points_all.mean(axis=0)
plothull(ax,
points_all_mean_time[..., :2],
griddims=curvs_map.shape,
vecs=mvecs,
c='k',
lw=0)
picturesave('fig.%s.%s' % ('curvature', sn),
work.plotdir,
backup=False,
version=True,
meta={})
plt.close()
def plot_curvature_basic(sn, **kwargs):
"""
Plot a map of the curvature.
Note: you must use a bin spacing of 2.0 or 5.0 and not 0.5 otherwise the curvature ends up biased
very strongly in one direction (this is a methodology error of some kind, related to binnning.)
A Gaussian blur is necessary to get usable information from the map, however this washes out the
curvature magnitudes leaving only a relative pattern.
"""
dat = data[sn]['data']
nframes = int(dat['nframes'])
def get(mn, fr, name):
return dat['%d.%d.%s' % (mn, fr, name)]
spacing = kwargs.pop('spacing', 2.0)
sigma = kwargs.pop('blur', 0.0)
if kwargs: raise Exception('unprocessed kwargs %s' % kwargs)
mvecs = np.mean([get(0, fr, 'vec') for fr in range(nframes)], axis=0)
ngrid = np.round(mvecs / spacing).astype(int)
#---curvature of each leaflet
curvs_mean_all = []
curvs_map_all = [np.zeros((ngrid[0], ngrid[1])) for mn in range(2)]
for mn in range(2)[::-1]:
curvs = np.zeros((ngrid[0], ngrid[1]))
curvs_counts = np.zeros((ngrid[0], ngrid[1]))
nmol = int(dat['%d.1.nmol' % mn])
curvs_mean = []
for fr in range(nframes):
simps = get(mn, fr, 'simplices')
pts = get(mn, fr, 'points')
vec = get(mn, fr, 'vec')
curv = get(mn, fr, 'mean')
#---previously implemented a filter
if False:
filt = np.where(np.abs(curv) < 0.5)
pts, curv = pts[filt], curv[filt]
#---repack the points in the box
pts_repack = (pts - np.floor(pts / vec) * vec)
pts_rounded = (pts_repack / (vec / ngrid)).astype(int)[:, :2]
curvs[pts_rounded[:nmol, 0], pts_rounded[:nmol, 1]] += curv[:nmol]
curvs_mean.append(curv[:nmol].mean())
curvs_counts[pts_rounded[:nmol, 0], pts_rounded[:nmol, 1]] += 1
curvs_mean_all.append(curvs_mean)
obs = np.where(curvs_counts > 0)
means = curvs[obs] / curvs_counts[obs]
curvs_map_all[mn][obs] = means
#---average the leaflets
curvs_map = np.mean(curvs_map_all, axis=0)
#---! removed the following line which sets nan (white on imshow). not sure what it is for
#---ironically back to nan for the imshow
if False: curvs_map[np.where(curvs_counts == 0)] = np.nan
#---! it might be nice to smooth the plots but beware!
if True:
curvs_map = scipy.ndimage.gaussian_filter(curvs_map,
sigma=(sigma, sigma),
order=0)
vmax = max([abs(j) for j in [curvs_map.max(), curvs_map.min()]])
ax = plt.gca()
im = ax.imshow(curvs_map,
cmap=mpl.cm.RdBu_r,
vmax=vmax,
vmin=-1 * vmax,
origin='lower',
extent=[0, mvecs[0], 0, mvecs[1]])
axins = inset_axes(ax,
width="5%",
height="100%",
loc=3,
bbox_to_anchor=(1.05, 0., 1., 1.),
bbox_transform=ax.transAxes,
borderpad=0)
cbar = plt.colorbar(im, cax=axins, orientation="vertical")
cbar.set_label(r'$C_0\,({nm}^{-1})$', labelpad=-40, y=1.05, rotation=0)
ax.set_xlabel('x (nm)')
ax.set_ylabel('y (nm)')
ax.tick_params(axis='y',
which='both',
left='off',
right='off',
labelleft='on')
ax.tick_params(axis='x',
which='both',
top='off',
bottom='off',
labelbottom='on')
axins.tick_params(axis='y',
which='both',
left='off',
right='off',
labelright='on')
ax.set_title('curvature map')
if sn in data_prot:
from render.wavevids import plothull
points_all = data_prot[sn]['data']['points_all']
points_all_mean_time = points_all.mean(axis=0)
plothull(ax,
points_all_mean_time[..., :2],
griddims=curvs_map.shape,
vecs=mvecs,
c='k',
lw=0)
picturesave('fig.%s.%s' % ('curvature', sn),
work.plotdir,
backup=False,
version=True,
meta={})
plt.close()
def individual_reviews_plotter(viewnames,
out_fn,
seep=None,
figsize=(10, 10),
horizontal=False,
wspace=0.4,
**kwargs):
"""
Loop over all upstream curvature-undulation coupling calculations and plot a panel of review plots.
"""
#---seeping namespace
if seep != None:
for key, val in seep.items():
globals()[key] = val
for tag in datas:
for sn in work.sns():
status('reviewing curvature for %s' % sn, tag='plot')
cmap_name = 'RdBu_r'
square_tiles_args = {}
if wspace != None: square_tiles_args.update(wspace=wspace)
axes, fig = square_tiles(len(viewnames),
figsize,
hspace=0.4,
favor_rows=horizontal,
**square_tiles_args)
#---several plots use the same data
vecs = postdat[sn]['vecs']
if len(vecs.shape) == 2: vecs = vecs.mean(axis=0)
griddims = datas[tag][sn]['cf'].shape
#---shared variables for several plots
cmax = np.abs(datas[tag][sn]['cf']).max()
cmax_instant = np.abs(datas[tag][sn]['cf_first']).max()
hicut = calcs[tag][sn]['calcs']['specs'].get('fitting', {}).get(
'high_cutoff', 1.0)
#---PLOT the mean curvature field
if 'average_field' in viewnames:
ax = axes[viewnames.index('average_field')]
im = ax.imshow(datas[tag][sn]['cf'].T,
origin='lower',
interpolation='nearest',
vmax=cmax,
vmin=-1 * cmax,
cmap=mpl.cm.__dict__[cmap_name],
extent=[0, vecs[0], 0, vecs[1]])
mean_trial = datas[tag][sn]['drop_gaussians_points'].transpose(
1, 0, 2).mean(axis=0)
ax.scatter(*mean_trial.T, s=1, c='k')
ax.set_title('average field')
add_colorbar(ax,
im,
title=r'$\mathrm{\langle C_0(x,y) \rangle}$')
#---PLOT a single instance of the neighborhood (good for debugging)
if 'neighborhood_static' in viewnames:
ax = axes[viewnames.index('neighborhood_static')]
debug_frame = 0
plot_hull_and_trial_centers(datas[tag],
sn,
ax,
debug_frame=debug_frame,
color='k')
ax.set_title('neighborhood, static, frame %d' % debug_frame)
#---PLOT a composite of several frames of the neighborhood to see the dynamics
if 'neighborhood_dynamic' in viewnames:
ax = axes[viewnames.index('neighborhood_dynamic')]
plot_hull_and_trial_centers(datas[tag], sn, ax, n_instances=10)
ax.set_title('neighborhood, dynamic')
#---PLOT the average height
if 'average_height' in viewnames:
mesh = data[undulations_name][sn]['data']['mesh']
surf = mesh.mean(axis=0).mean(axis=0)
surf -= surf.mean()
hmax = np.abs(surf).max()
ax = axes[viewnames.index('average_height')]
try:
im = ax.imshow(surf.T,
origin='lower',
interpolation='nearest',
cmap=mpl.cm.__dict__['RdBu_r'],
extent=[0, vecs[0], 0, vecs[1]],
vmax=hmax,
vmin=-1 * hmax)
except:
import ipdb
ipdb.set_trace()
try:
mean_prot_pts = [postdat[sn]['points_protein_mean'][:, :2]]
except:
try:
mean_prot_pts = data[protein_abstractor_name][sn][
'data']['points_all'].mean(axis=0)[..., :2]
except:
mean_prot_pts = [
data[protein_abstractor_name][sn]['data']
['points'].mean(axis=0)[:, :2]
]
plothull(ax,
mean_prot_pts,
griddims=datas[tag][sn]['cf'].shape,
vecs=vecs,
c='k',
lw=0)
ax.set_title('height profile')
ax.set_xlabel('x (nm)')
ax.set_ylabel('y (nm)')
add_colorbar(ax,
im,
title=r'$\mathrm{\langle z \rangle \, (nm)}$')
#---PLOT an example curvature field ('first' is hard-coded, instead of saving each frame)
if 'example_field' in viewnames:
ax = axes[viewnames.index('example_field')]
example_frame = 0
cf_first = datas[tag][sn]['cf_first']
im = ax.imshow(cf_first.T,
origin='lower',
interpolation='nearest',
vmax=cmax,
vmin=-1 * cmax,
cmap=mpl.cm.__dict__[cmap_name],
extent=[0, vecs[0], 0, vecs[1]])
add_colorbar(ax, im, title=r'$\mathrm{C_0\,({nm}^{-1})}$')
ax.set_xlim((0, vecs[0]))
ax.set_ylim((0, vecs[1]))
ax.set_xlabel('x (nm)')
ax.set_ylabel('y (nm)')
mean_trial = datas[tag][sn]['drop_gaussians_points'].transpose(
1, 0, 2)[example_frame]
ax.scatter(*mean_trial.T, s=1, c='k')
ax.set_title('curvature field', fontsize=10)
#---PLOT example curvature without the dots for the trial functions, including the protein hull
if 'example_field_no_neighborhood' in viewnames:
ax = axes[viewnames.index('example_field_no_neighborhood')]
example_frame = 0
cf_first = datas[tag][sn]['cf_first']
im = ax.imshow(cf_first.T,
origin='lower',
interpolation='nearest',
vmax=cmax,
vmin=-1 * cmax,
cmap=mpl.cm.__dict__[cmap_name],
extent=[0, vecs[0], 0, vecs[1]])
add_colorbar(ax, im, title=r'$\mathrm{C_0\,({nm}^{-1})}$')
ax.set_xlim((0, vecs[0]))
ax.set_ylim((0, vecs[1]))
ax.set_xlabel('x (nm)')
ax.set_ylabel('y (nm)')
mean_trial = datas[tag][sn]['drop_gaussians_points'].transpose(
1, 0, 2)[example_frame]
ax.set_title('curvature field', fontsize=10)
ex_prot_pts = data[protein_abstractor_name][sn]['data'][
'points'][example_frame][:, :2]
plothull(ax, [ex_prot_pts],
griddims=datas[tag][sn]['cf'].shape,
vecs=vecs,
c='k',
lw=0)
#---PLOT periodic view of the example field
if 'example_field_pbc' in viewnames:
ax = axes[viewnames.index('example_field_pbc')]
im = ax.imshow(
np.tile(cf_first.T, (3, 3)),
origin='lower',
interpolation='nearest',
vmax=cmax,
vmin=-1 * cmax,
cmap=mpl.cm.__dict__[cmap_name],
extent=[-vecs[0], 2 * vecs[0], -vecs[1], 2 * vecs[1]])
ax.set_title('curvature field (max %.3f)' % cmax_instant)
add_colorbar(ax, im, title=r'$\mathrm{C_0\,({nm}^{-1})}$')
#---PLOT periodic view of the average height
if 'average_height_pbc' in viewnames:
ax = axes[viewnames.index('average_height_pbc')]
im = ax.imshow(
np.tile(surf.T, (3, 3)),
origin='lower',
interpolation='nearest',
vmax=hmax,
vmin=-1 * hmax,
cmap=mpl.cm.__dict__[cmap_name],
extent=[-vecs[0], 2 * vecs[0], -vecs[1], 2 * vecs[1]])
ax.set_title('average height')
add_colorbar(ax, im, title=r'$\mathrm{\langle z \rangle}$')
#---track the protein
protein_pts = data[protein_abstractor_name][sn]['data'][
'points_all']
#### if protein_pts.shape[1]!=1: raise Exception('only one protein')
prot_traj = protein_pts[:, 0].mean(axis=1)[:, :2]
#### more proteins
prot_traj = np.concatenate(
data[protein_abstractor_name][sn]['data']
['points_all'].mean(axis=0)[..., :2])
for shift in np.concatenate(
np.transpose(np.meshgrid([-1, 0, 1],
[-1, 0, 1]))) * vecs[:2]:
ax.scatter(prot_traj[:, 0] + shift[0],
prot_traj[:, 1] + shift[1],
c='k',
s=1)
ax.set_xlim((-vecs[0], 2 * vecs[0]))
ax.set_ylim((-vecs[1], 2 * vecs[1]))
#---PLOT spectrum
if 'spectrum' in viewnames:
ax = axes[viewnames.index('spectrum')]
if len(datas[tag][sn]['qs_binned']) == len(
datas[tag][sn]['ratios']):
qs = datas[tag][sn]['qs_binned']
else:
datas[tag][sn]['qs']
ax.scatter(qs,
datas[tag][sn]['ratios'],
s=4,
c='k',
alpha=0.25)
#---! high cutoff is hard-coded here but needs to be removed to the yaml
#---! ...we need to get the default
band = qs <= hicut
ax.scatter(qs[band],
datas[tag][sn]['ratios'][band],
s=10,
c='k',
alpha=1.0)
ax.axhline(1.0, c='k')
ax.set_xscale('log')
ax.set_yscale('log')
error = datas[tag][sn]['bundle'][sn]['fun']
ax.set_title('full spectrum')
ax.grid(True)
ax.axvline(hicut, ymin=0.0, ymax=1.0, c='k', lw=2)
#---PLOT spectrum, relevant section
if 'spectrum_zoom' in viewnames:
ax = axes[viewnames.index('spectrum_zoom')]
qs = datas[tag][sn]['qs']
band = qs <= hicut
ys = datas[tag][sn]['ratios'][band]
ax.scatter(qs[band], ys, s=20, c='k', alpha=1.0, clip_on=False)
ax.axhline(1.0, c='k')
ax.set_xscale('log')
ax.set_yscale('log', subsy=[])
ax.set_yticks([min(ys), 1.0, max(ys)])
#---intransigent ticks!
ax.set_yticklabels([
('%.2f' % i if type(i) != bool else '')
for i in [min(ys), False, max(ys)]
])
ax.set_xlim(min(qs), hicut)
error = datas[tag][sn]['bundle'][sn]['fun']
ax.set_title('spectrum (error %.5f)' % error)
#---PLOT average height profile RELATIVE TO THE PROTEIN POSITION
if 'average_height_center' in viewnames:
mesh = data[undulations_name][sn]['data']['mesh']
surfs = mesh.mean(axis=0) #.mean(axis=0)
surfs_shifted = center_by_protein(sn, surfs)
surf = np.mean(surfs_shifted, axis=0)
surf -= surf.mean()
hmax = np.abs(surf).max()
ax = axes[viewnames.index('average_height_center')]
im = ax.imshow(surf.T,
origin='lower',
interpolation='nearest',
cmap=mpl.cm.__dict__['RdBu_r'],
extent=[0, vecs[0], 0, vecs[1]],
vmax=hmax,
vmin=-1 * hmax)
try:
mean_prot_pts = postdat[sn]['points_protein_mean'][:, :2]
except:
mean_prot_pts = data[protein_abstractor_name][sn]['data'][
'points'].mean(axis=0)[:, :2]
#---! still need to move the protein
if False:
plothull(ax, [mean_prot_pts],
griddims=datas[tag][sn]['cf'].shape,
vecs=vecs,
c='k',
lw=0)
plothull_protein_center_average(ax, sn, vecs, tag)
ax.set_title('height profile (centered)')
ax.set_xlabel('x (nm)')
ax.set_ylabel('y (nm)')
add_colorbar(ax,
im,
title=r'$\mathrm{\langle z \rangle \, (nm)}$')
#---PLOT the average field shifted as if the protein were in the center
#---we call this "naive" because it takes a single, average curvature field and moves it
#---...to the average position of the protein. if the protein crosses the periodic bounds,
#---...then moving *anything* to the average position of the protein makes no sense
#---...so see the explicit version below called "curvature_field_center"
if 'average_field_center' in viewnames:
ax = axes[viewnames.index('average_field_center')]
nframes = data[undulations_name][sn]['data']['mesh'].shape[1]
cf_centered = center_by_protein(sn,
np.array(
[datas[tag][sn]['cf']]),
static=True)
im = ax.imshow(cf_centered.T,
origin='lower',
interpolation='nearest',
vmax=cmax,
vmin=-1 * cmax,
cmap=mpl.cm.__dict__[cmap_name],
extent=[0, vecs[0], 0, vecs[1]])
mean_trial = datas[tag][sn]['drop_gaussians_points'].transpose(
1, 0, 2).mean(axis=0)
if False: ax.scatter(*mean_trial.T, s=1, c='k')
ax.set_title('curvature field (centered,naive)')
add_colorbar(ax, im, title=r'$\mathrm{C_0\,({nm}^{-1})}$')
#---PLOT the average curvature field, shifted to the instantaneous position of the protein
#---...and then averaged. this makes slightly more sense than "average_field_center" but we are
#---...still assuming an average and then just smearing it around. see "curvature_field_center"
if 'average_field_center_smear' in viewnames:
ax = axes[viewnames.index('average_field_center_smear')]
nframes = data[undulations_name][sn]['data']['mesh'].shape[1]
cf_centered = center_by_protein(
sn, np.tile(datas[tag][sn]['cf'],
(nframes, 1, 1))).mean(axis=0)
im = ax.imshow(cf_centered.T,
origin='lower',
interpolation='nearest',
vmax=cmax,
vmin=-1 * cmax,
cmap=mpl.cm.__dict__[cmap_name],
extent=[0, vecs[0], 0, vecs[1]])
mean_trial = datas[tag][sn]['drop_gaussians_points'].transpose(
1, 0, 2).mean(axis=0)
if False: ax.scatter(*mean_trial.T, s=1, c='k')
ax.set_title('curvature field (centered,smear)')
add_colorbar(ax, im, title=r'$\mathrm{C_0\,({nm}^{-1})}$')
#---PLOT the average of the protein-centered instantaneous curvature fields
#---note that you have to add "store_instantaneous_fields: True" to the design of the calculation
if 'curvature_field_center' in viewnames:
ax = axes[viewnames.index('curvature_field_center')]
nframes = data[undulations_name][sn]['data']['mesh'].shape[1]
cf_centered = center_by_protein(
sn, datas[tag][sn]['cfs']).mean(axis=0)
im = ax.imshow(cf_centered.T,
origin='lower',
interpolation='nearest',
vmax=cmax,
vmin=-1 * cmax,
cmap=mpl.cm.__dict__[cmap_name],
extent=[0, vecs[0], 0, vecs[1]])
mean_trial = datas[tag][sn]['drop_gaussians_points'].transpose(
1, 0, 2).mean(axis=0)
if False: ax.scatter(*mean_trial.T, s=1, c='k')
ax.set_title('curvature field (centered)')
ax.set_xlabel('x (nm)')
ax.set_ylabel('y (nm)')
add_colorbar(
ax,
im,
title=r'$\mathrm{\langle C_0 \rangle \,({nm}^{-1})$')
plothull_protein_center_average(ax, sn, vecs, tag)
#---no tick marks on anything
for ax in axes:
ax.tick_params(axis='x',
which='both',
bottom='off',
top='off',
labelbottom='on')
ax.tick_params(axis='y',
which='both',
left='off',
right='off',
labelbottom='off')
#---the metadata for this plot comes from the design section
try:
meta = calcs[tag][sn]['calcs']['specs']['specs']
#---! custom upstream calculations for e.g. dextran project put the specs one level down
except:
meta = calcs[tag][sn]['calcs']['specs']
#---add high cutoff (from fitting parameters if defined) to the meta
meta['high_cutoff'] = hicut
picturesave('fig.%s.%s' % (out_fn, sn),
work.plotdir,
backup=False,
version=True,
meta=meta)