Skip to content

Commit 3230b3d

Browse files
kjsandbrinkMMathisLab
kjsandbrink
authored and
MMathisLab
committed
cleaned up code
1 parent ca5b0d6 commit 3230b3d

20 files changed

Lines changed: 120 additions & 544 deletions

single_cell/combined_violinquantiles_controls.py renamed to single_cell/combined_violinquantiles.py

Lines changed: 23 additions & 56 deletions
Original file line numberDiff line numberDiff line change
@@ -17,16 +17,14 @@
1717

1818
from matplotlib.ticker import FormatStrFormatter
1919

20-
# def format_axis(ax):
21-
# ax.spines['top'].set_visible(False)
22-
# ax.spines['right'].set_visible(False)
23-
# ax.get_xaxis().tick_bottom()
24-
# ax.get_yaxis().tick_left()
25-
# ax.xaxis.set_tick_params(size=6)
26-
# ax.yaxis.set_tick_params(size=6)
27-
2820
## eLife
2921
def format_axis(ax):
22+
''' format axis to unified format
23+
24+
Arguments
25+
---------
26+
ax : matplotlib axis object
27+
'''
3028
ax.spines['top'].set_visible(False)
3129
ax.spines['right'].set_visible(False)
3230
ax.get_xaxis().tick_bottom()
@@ -161,6 +159,13 @@ def get_combined_modevals(model, runinfo):
161159
# %% COMPARISON VIOLIN PLOT
162160

163161
def clip(vp, lr):
162+
''' clips off half of a violin plot for split violin plots
163+
164+
Arguments
165+
---------
166+
vp : matplotlib violin plot object
167+
lr : str, one of ['l', 'r'] specifying whether we want to keep the left or right side of the violin plot
168+
'''
164169

165170
for b in vp['bodies']:
166171
m = np.mean(b.get_paths()[0].vertices[:, 0])
@@ -172,14 +177,17 @@ def clip(vp, lr):
172177

173178

174179
def plot_compvp(trainedmodevals, controlmodevals, trainedmodel, regcomp = False, modnames = modnames, ifsets_to_quantile = [0,3]):
175-
''' Plot the comparison violin plot showing the distribution of tuning strengths
180+
''' Plot the comparison violin plot showing the distribution of tuning strengths -- plotting format associated with original submission
176181
177182
Arguments
178183
---------
179184
trainedmodevals : list [nr layers] of np.arrays, store performance of each neuron in trained model
180185
controlmodevals : list [nr layers] of np.arrays, store performance of each neuron in control model
181186
trainedmodel : dict
182-
187+
regcomp : bool, True if we are comparing ART and TDT, False if we are comparing trained and random models
188+
modnames : list of str, list of variables/feature sets that we are plotting e.g. "vel", "acc" etc.
189+
ifsets_to_quantile : list of ints, indices of variables in mod_names for which to add the 90% quantile
190+
183191
Returns
184192
-------
185193
fig : plt.figure, comparison violin plot
@@ -198,7 +206,6 @@ def plot_compvp(trainedmodevals, controlmodevals, trainedmodel, regcomp = False,
198206
controlcmap = matplotlib.cm.get_cmap('Greys_r') #spatial_temporal
199207
else:
200208
controlcmap = matplotlib.cm.get_cmap(trainedmodel['regression_cmap'])
201-
#cmaps = [trainedcmap, controlcmap]
202209

203210
## SET OTHER PLOTTING VARIABLES
204211

@@ -218,7 +225,6 @@ def plot_compvp(trainedmodevals, controlmodevals, trainedmodel, regcomp = False,
218225
patches = []
219226

220227
ccolorindex = 3
221-
#for (modevals, cmap, alpha, zorder, lr) in zip([controlmodevals, trainedmodevals], [controlcmap, trainedcmap], [[0.8, 0.5], [0.8, 0.7]], [2,1], ['r', 'l']):
222228
for (modevals, cmap, alpha, zorder, lr) in zip([controlmodevals, trainedmodevals], [controlcmap, trainedcmap], [[1, 1], [1, 1]], [2,1], ['r', 'l']):
223229
for i, mod in enumerate(modevals):
224230

@@ -264,7 +270,6 @@ def plot_compvp(trainedmodevals, controlmodevals, trainedmodel, regcomp = False,
264270
print("empty array, can't do violin plot", e)
265271
vp = None
266272

267-
#patches.append(mpatches.Patch(color=cmap(cidx[i]), alpha=0.7))
268273
patches.append(mpatches.Patch(color=cmap(cidx[i]), alpha=1))
269274

270275
vps.append(vp)
@@ -299,14 +304,16 @@ def plot_compvp(trainedmodevals, controlmodevals, trainedmodel, regcomp = False,
299304

300305
return fig
301306

302-
def plot_compvp_v3(trainedmodevals, controlmodevals, trainedmodel, regcomp = False, modnames = modnames, ifsets_to_quantile = [0,3]):
303-
''' Plot the comparison violin plot showing the distribution of tuning strengths
307+
def plot_compvp_v3(trainedmodevals, controlmodevals, trainedmodel, regcomp = False, modnames = modnames):
308+
''' Plot the comparison violin plot showing the distribution of tuning strengths, plot format associated with eLife resubmission
304309
305310
Arguments
306311
---------
307312
trainedmodevals : list [nr layers] of np.arrays, store performance of each neuron in trained model
308313
controlmodevals : list [nr layers] of np.arrays, store performance of each neuron in control model
309314
trainedmodel : dict
315+
regcomp : bool, True if we are comparing ART and TDT, False if we are comparing trained and random models
316+
modnames : list of str, list of variables/feature sets that we are plotting e.g. "vel", "acc" etc.
310317
311318
Returns
312319
-------
@@ -319,7 +326,6 @@ def plot_compvp_v3(trainedmodevals, controlmodevals, trainedmodel, regcomp = Fal
319326

320327
space = 0.8
321328
width = 0.6
322-
#lspace = space*nmods + 1
323329
lspace = space*nmods + 1.6
324330

325331
trainedcmap = matplotlib.cm.get_cmap(trainedmodel['cmap']) #spatial_temporal
@@ -328,7 +334,6 @@ def plot_compvp_v3(trainedmodevals, controlmodevals, trainedmodel, regcomp = Fal
328334
controlcmap = matplotlib.cm.get_cmap('Greys_r') #spatial_temporal
329335
else:
330336
controlcmap = matplotlib.cm.get_cmap(trainedmodel['regression_cmap'])
331-
#cmaps = [trainedcmap, controlcmap]
332337

333338
## SET OTHER PLOTTING VARIABLES
334339

@@ -347,14 +352,10 @@ def plot_compvp_v3(trainedmodevals, controlmodevals, trainedmodel, regcomp = Fal
347352
patches = []
348353

349354
ccolorindex = 3
350-
#for (modevals, cmap, alpha, zorder, lr) in zip([controlmodevals, trainedmodevals], [controlcmap, trainedcmap], [[0.8, 0.5], [0.8, 0.7]], [2,1], ['r', 'l']):
351355
for (modevals, cmap, alpha, zorder, lr) in zip([controlmodevals, trainedmodevals], [controlcmap, trainedcmap], [[1, 1], [1, 1]], [2,1], ['r', 'l']):
352356
for i, mod in enumerate(modevals):
353357

354-
#print(mod)
355358
mod = [x.reshape((-1,)) for x in mod]
356-
#for x in mod:
357-
# print(x.shape)
358359

359360
##exclude r2 == 1 scores and r2 < -0.1
360361
mod = [x[(x != 1) & (x > -0.1)] for x in mod]
@@ -393,47 +394,14 @@ def plot_compvp_v3(trainedmodevals, controlmodevals, trainedmodel, regcomp = Fal
393394
print("empty array, can't do violin plot", e)
394395
vp = None
395396
patches.append(mpatches.Patch(color=cmap(cidx[i]), alpha=1))
396-
#patches.append(mpatches.Patch(color=cmap(cidx[i]), alpha=0.7))
397397
#patches.append(mpatches.Patch(color=matplotlib.cm.get_cmap('Greys_r') (cidx[i]), alpha=0.7)) ##eLife
398398

399399

400400
vps.append(vp)
401-
402-
# #Quantiles
403-
# q = 0.9
404-
# marker = 's'
405-
406-
# for i in ifsets_to_quantile:
407-
# mod = modevals[i]
408-
409-
# mod = [x[x != 1] for x in mod]
410-
411-
# q90s = np.zeros((nlayers,))
412-
413-
# for ilayer, layer in enumerate(mod):
414-
# try:
415-
# q90s[ilayer] = np.quantile(layer, q)
416-
# except e:
417-
# print(e)
418-
# ax1.plot([ilayer*lspace+space*i+1 for ilayer in range(nlayers)],
419-
# q90s, color=cmap(cidx[i]), marker=marker, alpha=alpha[0])
420-
421401

422402
format_axis(plt.gca())
423403
format_axis(ax1)
424404

425-
## SET AXIS WIDTHS
426-
#for axis in ['top','bottom','left','right']:
427-
# ax1.spines[axis].set_linewidth(1.5)
428-
429-
# increase tick width
430-
#ax1.tick_params(width=1.5, labels=18)
431-
432-
433-
#ax1.yaxis.label.set_size(18)
434-
#ax1.xaxis.label.set_size(18)
435-
436-
437405
leg = plt.legend(patches[5:], modnames, loc='upper right')
438406
ax1.add_artist(leg)
439407
if not regcomp:
@@ -491,6 +459,7 @@ def plot_compvp_ee(trainedmodevals, controlmodevals, trainedmodel, regcomp = Fal
491459
trainedmodevals : list [nr layers] of np.arrays, store performance of each neuron in trained model
492460
controlmodevals : list [nr layers] of np.arrays, store performance of each neuron in control model
493461
trainedmodel : dict
462+
regcomp : bool, True if we are comparing ART and TDT, False if we are comparing trained and random models
494463
495464
Returns
496465
-------
@@ -531,7 +500,6 @@ def plot_compvp_ee(trainedmodevals, controlmodevals, trainedmodel, regcomp = Fal
531500
patches = []
532501

533502
ccolorindex = 1
534-
#for (modevals, cmap, alpha, zorder, lr) in zip([controlmodevals, trainedmodevals], [controlcmap, trainedcmap], [[0.8, 0.5], [0.8, 0.7]], [2,1], ['r', 'l']):
535503
for (modevals, cmap, alpha, zorder, lr) in zip([controlmodevals, trainedmodevals], [controlcmap, trainedcmap], [[1, 1], [1, 1]], [2,1], ['r', 'l']):
536504
for i, mod in enumerate(modevals):
537505

@@ -574,7 +542,6 @@ def plot_compvp_ee(trainedmodevals, controlmodevals, trainedmodel, regcomp = Fal
574542
vp = None
575543

576544

577-
#patches.append(mpatches.Patch(color=cmap(cidx[i]), alpha=0.7))
578545
patches.append(mpatches.Patch(color=cmap(cidx[i]), alpha=1))
579546

580547
vps.append(vp)

0 commit comments

Comments
 (0)