This notebook is part of smFRET burst analysis software FRETBursts.
In this notebook shows how to plot different styles of μs-ALEX histograms and $E$ and $S$ marginal distributions. For a complete tutorial on burst analysis see FRETBursts - us-ALEX smFRET burst analysis.
In [ ]:
from fretbursts import *
In [ ]:
sns = init_notebook(apionly=True)
print('seaborn version: ', sns.__version__)
In [ ]:
# Tweak here matplotlib style
import matplotlib as mpl
mpl.rcParams['font.sans-serif'].insert(0, 'Arial')
mpl.rcParams['font.size'] = 12
%config InlineBackend.figure_format = 'retina'
In [ ]:
url = 'http://files.figshare.com/2182601/0023uLRpitc_NTP_20dT_0.5GndCl.hdf5'
download_file(url, save_dir='./data')
full_fname = "./data/0023uLRpitc_NTP_20dT_0.5GndCl.hdf5"
d = loader.photon_hdf5(full_fname)
loader.alex_apply_period(d)
d.calc_bg(bg.exp_fit, time_s=1000, tail_min_us=(800, 4000, 1500, 1000, 3000))
d.burst_search(L=10, m=10, F=6)
ds = d.select_bursts(select_bursts.size, add_naa=True, th1=30)
In [ ]:
alex_jointplot(ds)
The inner plot in an hexbin plot, basically a 2D histogram with hexagonal bins. This kind of histograms resembles a scatter plot when sample size is small, and is immune from grid artifacts typical of rectangular grids. For more info for hexbin see this document.
The marginal plots are histograms with an overlay KDE plot. The same FRETBursts function that plots standalone E and S histograms is used here to plot the marginals in the joint plot.
Below I show how to customize appearance and type of this plot.
By default the colormap range is computed on the range S=[0.2, 0.8], so that the FRET populations (S ~ 0.5) have more contrast.
To normalize the colormap to the whole data use the vmax argument:
In [ ]:
alex_jointplot(ds, vmax_fret=False)
In [ ]:
alex_jointplot(ds, vmax_fret=False, marginal_color=8)
In [ ]:
alex_jointplot(ds, vmax_fret=False, marginal_color=7)
In [ ]:
alex_jointplot(ds, kind='kde')
Or you can manually choose the max value mapped by the colormap (vmax):
In [ ]:
alex_jointplot(ds, vmax=40)
Changing the colormap will affect both inner and marginal plots:
In [ ]:
alex_jointplot(ds, cmap='plasma')
To pick a different color from the colormap for the marginal histograms use histcolor_id:
In [ ]:
alex_jointplot(ds, cmap='plasma', marginal_color=83)
The inner plot can be changed to a scatter plot or a KDE plot:
In [ ]:
alex_jointplot(ds, kind='scatter')
In [ ]:
alex_jointplot(ds, kind='kde')
In [ ]:
dsf = ds.select_bursts(select_bursts.naa, th1=40)
alex_jointplot(dsf, kind='kde',
joint_kws={'shade': False, 'n_levels': 12, 'bw': 0.04})
In [ ]:
plt.figure(figsize=(5,5))
hexbin_alex(ds)
You can get an handle of the different axes in the figure for layout customization:
In [ ]:
g = alex_jointplot(ds)
g.ax_marg_x.grid(False)
g.ax_marg_y.grid(False)
g.ax_joint.set_xlim(-0.1, 1.1)
g.ax_joint.set_ylim(-0.1, 1.1)
alex_jointplot returns g which contains the axis handles (g.ax_join, g.ax_marg_x, g.ax_marg_y).
The object g is a seaborn.JointGrid.
In [ ]:
g = alex_jointplot(ds)
g.ax_marg_x.grid(False)
g.ax_marg_y.grid(False)
g.ax_joint.set_xlim(-0.19, 1.19)
g.ax_joint.set_ylim(-0.19, 1.19)
plt.subplots_adjust(wspace=0, hspace=0)
g.ax_marg_y.spines['bottom'].set_visible(True)
g.ax_marg_x.spines['left'].set_visible(True)
g.ax_marg_y.tick_params(reset=True, bottom=True, top=False, right=False, labelleft=False)
g.ax_marg_x.tick_params(reset=True, left=True, top=False, right=False, labelbottom=False)
In [ ]:
g = alex_jointplot(ds)
g.ax_marg_x.grid(False)
g.ax_marg_y.grid(False)
g.ax_joint.set_xlim(-0.19, 1.19)
g.ax_joint.set_ylim(-0.19, 1.19)
plt.subplots_adjust(wspace=0, hspace=0)
g.ax_marg_y.tick_params(reset=True, bottom=True, top=False, right=False, labelleft=False)
g.ax_marg_x.tick_params(reset=True, left=True, top=False, right=False, labelbottom=False)
In [ ]:
g = alex_jointplot(ds)
g.ax_marg_x.grid(False, axis='x')
g.ax_marg_y.grid(False, axis='y')
g.ax_joint.set_xlim(-0.19, 1.19)
g.ax_joint.set_ylim(-0.19, 1.19)
plt.subplots_adjust(wspace=0, hspace=0)
Additional arguments can be passed to the inner or marginal plots passing
a dictionary to joint_kws and marginal_kws respectively.
The marginal plots are created by hist_burst_data
which is the same function used to plot standalone E and S histograms
in FRETBursts.
For example, we can remove the KDE overlay like this:
In [ ]:
alex_jointplot(ds, marginal_kws={'show_kde': False})
In [ ]:
from ipywidgets import widgets, interact, interactive, fixed
from IPython.display import display, display_png, display_svg, clear_output
from IPython.core.pylabtools import print_figure
In [ ]:
cmaps = ['viridis', 'plasma', 'inferno', 'magma',
'afmhot', 'Blues', 'BuGn', 'BuPu', 'GnBu', 'YlGnBu',
'coolwarm', 'RdYlBu', 'RdYlGn', 'Spectral',]# 'icefire'] uncomment if using seaborn 0.8
In [ ]:
@interact(overlay = widgets.RadioButtons(options=['fit model', 'KDE'], value='KDE'),
binwidth = widgets.FloatText(value=0.03, min=0.01, max=1),
bandwidth = widgets.FloatText(value=0.03, min=0.01, max=1),
gridsize = (10, 100),
min_size=(10, 500, 5),
cmap=widgets.Dropdown(value='Spectral', options=cmaps),
reverse_cmap = True,
vmax_fret = True,
)
def plot_(min_size=50, overlay='KDE', binwidth=0.03, bandwidth=0.03,
gridsize=50, cmap='Spectral', reverse_cmap=False,
vmax_fret=True):
dx = d.select_bursts(select_bursts.size, add_naa=True, th1=min_size)
bext.bursts_fitter(dx, 'E', binwidth=binwidth, bandwidth=bandwidth,
model=mfit.factory_three_gaussians())
bext.bursts_fitter(dx, 'S', binwidth=binwidth, bandwidth=bandwidth,
model=mfit.factory_two_gaussians())
if reverse_cmap: cmap += '_r'
if binwidth < 0.01: binwidth = 0.01
if bandwidth < 0.01: bandwidth = 0.01
if overlay == 'fit model':
marginal_kws = dict(binwidth=binwidth, show_model=True, pdf=True,
show_kde=False)
else:
marginal_kws = dict(binwidth=binwidth, show_kde=True,
bandwidth=bandwidth)
alex_jointplot(dx, cmap=cmap, gridsize=gridsize, vmax_fret=vmax_fret,
marginal_kws=marginal_kws,)
fig = gcf()
plt.close()
display(fig)