将Matplotlib注释与颜色栏对齐

Question

I'd like to plot a colorbar, with a string(sequence) aligned to the bottom of it. I have a working example posted below, but it's not really an ideal solution because there are a few hardcoded values. Are there any other/better/smarter ways that I could do this? Accommodating various font sizes dynamically would be ideal.

requirements:

1. The text must be selectable in a web browser as one continuous line (svg format in webbrowsers allows us to click and drag select matplotlib text)
   • I'd like to select the sections of the text. ie, if i see a particularly interesting region based on my scorelist, i'd like to select that sections and copy it to my clipboard
2. The text must align perfectly to the colorbar behind it.

If you were to make each letter an axes tick, they would no longer be selectable as a single continuous string (actually a newline separates them) which I don't want.

import numpy as np
import matplotlib.pyplot as plt
from numpy import array
from mpl_toolkits.axes_grid1 import make_axes_locatable


myseq = "AERGERGEREKLKLKLLLKKLWEWRKLEWRLWKERLKKLYKLTELWLKLRWELKHLW"
scorelist = [np.random.randint(0,5) for x in myseq ]

# SET FIGURE SIZE, Not ideal, but sort of works
fig = plt.figure(figsize=(len(myseq)/6.64175,5)) # MAGIC NUMBER?
ax = plt.subplot()
scorearray = array([scorelist,scorelist])

# Plotting the colorbar
cmap = plt.cm.rainbow
im = ax.imshow(scorearray, extent=[-0.5, len(scorelist)-0.5, 0, 5], cmap=cmap)
divider = make_axes_locatable(ax)
cax = divider.append_axes("top", size="5%", pad=0.04)
cbar = plt.colorbar(im, cax=cax, orientation='horizontal')
cbar.ax.xaxis.set_ticks_position('top')
ax.yaxis.set_visible(False)
ax.text(-0.5,-1.1,myseq, size=14, name='Courier new')

# Use 1 indexing for human counting
ax.set_xticks([0] + list(np.arange(9, len(myseq), 10)));
ax.set_xticklabels(['1'] + [str(i) for i in range(10, len(myseq), 10)]);
ax.set_xticks(np.arange(-0.5, len(myseq), 1), minor=True);

ax.grid(which='minor', color='w', linestyle='-', linewidth=2)
plt.rcParams['svg.fonttype'] = 'none'
# plt.show()
plt.savefig("alignment.svg")

Basically I want to be able to look at colorbar representation of the scorelist, and manually select regions that I find interesting, and require the scorelist to align to my on-graph text nicely.

Answer 1

Note that the following does not actually produce the correct svg output. One may still use it for png output though.

An option is to calculate the needed figure width according to the given text. So the idea is to first create the text with some font and size, then "measure" how large it is and according to that number set the axes width. Now the axes width itself is not directly available, but rather subject to the figure width and the margins being set, according to figwidth = axes width/(margin_right - margin_left) . We would then set the text width as the axes width. The text itself is positionned in the middle of the axes with center alignment, such that once the axes has the same width than the text they (almost) perfectly align.

import numpy as np
import matplotlib.pyplot as plt
from mpl_toolkits.axes_grid1 import make_axes_locatable

#### free input parameters ####
figheight = 4 # inch
subplotpars = dict(left=0.1, right=0.9)
fontsize = 16
fontname = 'Courier new' # need to choose a monospace font here

myseq = "AERGERGEREKLKLKLLLKKLWEWRKLEWRLWKERLKKLYKLTELWLKLRWELKHLW"
scorelist = [np.random.randint(0,5) for x in myseq ]

#### automatic plotting ####
fig = plt.figure(figsize=(6,figheight)) # at this point only height will matter
fig.subplots_adjust(**subplotpars)
ax = plt.subplot()
scorearray = np.array([scorelist,scorelist])

# Plotting the colorbar
cmap = plt.cm.rainbow
im = ax.imshow(scorearray, extent=[-0.5, len(scorelist)-0.5, 0, 5], cmap=cmap)
divider = make_axes_locatable(ax)
cax = divider.append_axes("top", size="5%", pad=0.04)
cbar = plt.colorbar(im, cax=cax, orientation='horizontal')
cbar.ax.xaxis.set_ticks_position('top')
ax.yaxis.set_visible(False)

# position text centered at middle of axes
text = ax.text(0.5,-0.08, myseq, size=fontsize, name=fontname, 
               transform=ax.transAxes, va="top", ha="center")
# draw canvas to fix positions of elements
fig.canvas.draw()
bbox = text.get_window_extent()
# calculate figure width according to width of text
figwidth=bbox.width/float(fig.dpi)/(subplotpars["right"]-subplotpars["left"])
fig.set_size_inches((figwidth, fig.get_size_inches()[1]))


ax.set_xticks([0] + list(np.arange(9, len(myseq), 10)));
ax.set_xticklabels(['1'] + [str(i) for i in range(10, len(myseq), 10)]);
ax.set_xticks(np.arange(-0.5, len(myseq), 1), minor=True);
ax.xaxis.set_tick_params(pad=10)

ax.grid(which='minor', color='w', linestyle='-', linewidth=2)
plt.rcParams['svg.fonttype'] = 'none'
plt.show()

Now it's relatively easy to play with the free input parameters and observe the changes.
Note that in principle one could also automatically determine the padding between axes and text (here hardcoded to 0.08) and between axes and ticklabels (here hardcoded to 10 points); but as I don't know if that is needed I left that out.