def plot(self):
clrs = {0: 'y', 1: 'r', 2: 'g', 3: 'b'}
# fig = plt.figure()
fix, ax = plt.subplots()
if self.is_valid_cell():
# Get bbox and translate to origin
rects = []
bbox = self.cell.bbox
new_bbox = list(
((bbox[0][0] - bbox[0][0], bbox[0][1] - bbox[0][1]),
(bbox[1][0] - bbox[0][0], bbox[1][1] - bbox[0][1])))
# create grid based on bbox max/min in appropriate steps
for obj in self.cell.objects:
rect = Rectangle(obj.points[0],
self.get_width(obj.points),
self.get_height(obj.points),
alpha=0.4,
color=clrs[obj.num])
rect.set_label(obj.name)
rects.append(rect)
# plotting stuff...
# do stuff...
collection = PatchCollection(rects, match_original=True)
ax.add_collection(collection)
ax.legend(loc='lower center', handles=rects)
# plt.plot(rects)
plt.show()
else:
print("Invalid Cell")
def __init__(self, df, state = np):
#select a random ROI
roi = df.ROI_ID.unique()[state]
df = df[df.ROI_ID == roi]
df_on_f = df.dF_on_F.values[:self.total_points]
x = np.linspace(0,len(df_on_f)//5, len(df_on_f))
nontrial_idxs = pd.isnull(df.Trial_ID.values)[:self.total_points]
trial_idxs = ~nontrial_idxs
trial_starts, = np.array(np.where(rising_edges(trial_idxs,cutoff=0.5)))//5
trial_ends, = np.array(np.where(falling_edges(trial_idxs,cutoff=0.5)))//5
self.fig = plt.figure(figsize=(10,11), tight_layout=True)
gs = GridSpec(self.nrows, 8, self.fig)
ax = np.zeros(self.nplots,dtype=object)
ax[0] = self.fig.add_subplot(gs[0,:])
ax[1] = self.fig.add_subplot(gs[1,:])
ax[2] = self.fig.add_subplot(gs[2,:-2])
ax[3] = self.fig.add_subplot(gs[2,-2:])
ax[4] = self.fig.add_subplot(gs[3,0:2])
ax[5] = self.fig.add_subplot(gs[3,3:5])
ax[6] = self.fig.add_subplot(gs[3,6:8])
ax[7] = self.fig.add_subplot(gs[4,:])
minus_plot = self.fig.add_subplot(gs[3,2])
minus_plot.set_axis_off(); minus_plot.text(0.5,0.5,"$\\bf{-}$",
transform=minus_plot.transAxes)
equals_plot = self.fig.add_subplot(gs[3,5])
equals_plot.set_axis_off(); equals_plot.text(0.5,0.5,"$\\bf{=}$",
transform=equals_plot.transAxes)
dff_trial = df_on_f.copy(); dff_trial[nontrial_idxs]=np.nan
dff_nontrial = df_on_f.copy(); dff_nontrial[trial_idxs]=np.nan
ax[0].plot(x,dff_nontrial, color='k', label = "Fluorescnece outside trials")
ax[0].plot(x,dff_trial,color=sns.color_palette()[3],
label="Fluorescence during trials (discarded)")
ymax,ymin = ax[0].get_ylim()
for start,end in zip(trial_starts,trial_ends):
xy = (start,ymin)
width = end-start
height = ymax - ymin
patch = Rectangle(xy,width,height, alpha=0.6)
ax[0].add_patch(patch)
patch.set_label("Trials")
ax[1].plot(x,dff_nontrial, color = 'k')
for a in ax[:2]:
a.set_xlim(0,self.total_points//5)
ymin,ymax = ax[1].get_ylim()
xmin,xmax = ax[1].get_xlim()
training_dataset = Rectangle((xmin,ymin), 0.8*(xmax-xmin),
(ymax-ymin),
color=sns.color_palette()[1],
label = "Training Dataset",
alpha = 0.6)
testing_dataset = Rectangle(((xmin+ 0.8*(xmax-xmin)),ymin),
0.2*(xmax-xmin),
(ymax-ymin),
color=sns.color_palette()[2],
label = "Testing Dataset",
alpha = 0.6)
for patch in (training_dataset,testing_dataset):
ax[1].add_patch(patch)
training_x = x[:self.training_points]
testing_x = x[self.training_points:self.total_points]
training_dat = dff_nontrial[:self.training_points]
training_licking = df.lick_factor[:self.training_points].values
testing_dat = dff_nontrial[self.training_points:self.total_points]
testing_licking = df.lick_factor[self.training_points:self.total_points]
#define some R variables
r.globalenv["dff"] = r.FloatVector(dff_nontrial)
r.globalenv["licking"]= r.FloatVector(df.lick_factor[:self.total_points].values)
r.r("dat <- data.frame(dff,licking)")
r.r(f"train <- dat[1:{self.training_points},]")
r.r(f"test <- dat[-(1:{self.training_points}),]")
r.r("model <- lm(dff~as.factor(licking),na.action=na.omit,dat=train)")
r.r("preds <- predict(model,test)")
model = r.r.model
fitted_vals = np.array(r.r("predict(model,train)"))
fitted_vals[trial_idxs[:self.training_points]] = np.nan
preds = np.array(r.r.preds)
preds[trial_idxs[self.training_points:self.total_points]] = np.nan
ax[2].plot(training_x,training_dat,
color = 'k')
ax[2].plot(training_x,fitted_vals, color=sns.color_palette()[1],
label = "Fitted model values based on licking (significant)")
ax[3].plot(x[self.total_points*4//5:self.total_points],dff_nontrial[self.total_points*4//5:self.total_points],
color = 'k')
ax[3].plot(testing_x,preds, color = sns.color_palette()[2],
label = "Model predictions")
ax[3].set_yticklabels([])
ax[4].plot(testing_x,testing_dat,color='k',label="Testing data")
ax[5].plot(testing_x,preds, color=sns.color_palette()[2],label="Prediction")
ax[6].plot(testing_x,testing_dat - preds, color = 'k',label="Licking-adjusted fluorescence")
for idx,a in enumerate(ax):
if idx in (0,1,2,3,4,7):
a.set_ylabel("Fluorescence ($\\Delta$F/F0 units)")
ymin,ymax = (min(a.get_ylim()[0] for a in ax),
max(a.get_ylim()[1] for a in ax))
for a in ax[-3:-1]:
a.set_yticklabels([])
ax[-1].plot(testing_x,testing_dat-preds, color='k',
label = "Licking-adjusted fluroescence")
r.r("res <- test")
r.r("res$dff = res$dff - preds")
r.r("model2 <- lm(dff~as.factor(licking),na.action=na.omit,dat=res)")
newpreds = np.array(r.r("predict(model2,test)"))
newpreds[trial_idxs[self.training_points:self.total_points]] = np.nan
ax[-1].plot(testing_x,newpreds,color=sns.color_palette()[3],
label = "A new fitted model, based on licking (NOT significant))")
for a in ax:
a.set_ylim((ymin,ymax))
a.legend(loc="upper right")
a.set_xticklabels([])
ax[3].set_ylabel("")
for a,char in zip(ax,"ABCDEFGE"):
a.set_title("$\\bf{(" + f"{char}" +")}$", loc="right")
def __init__(self,df,state=0):
fig,ax = plt.subplots(nrows = 5, figsize = (10,11), tight_layout=True)
roi = df.sample(random_state=state).ROI_ID.values[0]
df = df[df.ROI_ID == roi]
df_on_f = df.dF_on_F.values[:self.npoints]
x = np.linspace(0,len(df_on_f)//5, len(df_on_f))
nontrial_idxs = pd.isnull(df.Trial_ID.values)[:self.npoints]
trial_idxs = ~nontrial_idxs
trial_starts, = np.array(np.where(rising_edges(trial_idxs,cutoff=0.5)))//5
trial_ends, = np.array(np.where(falling_edges(trial_idxs,cutoff=0.5)))//5
dff_trial = df_on_f.copy(); dff_trial[nontrial_idxs]=np.nan
dff_nontrial = df_on_f.copy(); dff_nontrial[trial_idxs]=np.nan
ax[0].plot(x,dff_trial,color=sns.color_palette('bright')[0],
label="Fluorescence during trials")
ax[0].plot(x,dff_nontrial, color=sns.color_palette()[1],
label = "Fluorescnece outside trials")
ymax,ymin = ax[0].get_ylim()
for start,end in zip(trial_starts,trial_ends):
xy = (start,ymin)
width = end-start
height = ymax - ymin
patch = Rectangle(xy,width,height, alpha=0.6)
ax[0].add_patch(patch)
r.globalenv["dff"] = r.FloatVector(dff_nontrial)
r.globalenv["licking"]= r.FloatVector(df.lick_factor[:self.npoints].values)
r.r("nontrial <- data.frame(dff,licking)")
r.globalenv["dff"] = r.FloatVector(dff_trial)
r.globalenv["licking"]= r.FloatVector(df.lick_factor[:self.npoints].values)
r.r("trial <- data.frame(dff,licking)")
r.r("model <- lm(dff~as.factor(licking),na.action=na.omit,dat=nontrial)")
r.r("preds <- predict(model,trial)")
patch.set_label("Trials")
ax[1].plot(x,dff_nontrial, color = sns.color_palette()[1],label="Fluorescence outside trials")
ax[1].plot(x,np.array(r.r.preds),color=sns.color_palette()[3],
label="Fitted licking model")
ax[2].plot(x,dff_trial,color=sns.color_palette('bright')[0])
ax[2].plot(x,np.array(r.r.preds),color=sns.color_palette()[3],
label = "Model predictions")
ax[3].plot(x,(dff_trial-np.array(r.r.preds)),
color = sns.color_palette('bright')[0],
label = "Licking-corrected trial fluorescence")
for start,end in zip(trial_starts,trial_ends):
xy = (start,ymin)
width = end-start
height = ymax - ymin
patch1 = Rectangle(xy,width,height, alpha=0.3)
patch2 = Rectangle(xy,width,height, alpha=0.3)
ax[3].add_patch(patch1)
ax[4].add_patch(patch2)
trials = dff_trial[~np.isnan(dff_trial)].reshape(-1,25)
tone = trials[:,:5].mean(axis=-1)
stim = trials[:,5:15].mean(axis=-1)
resp = trials[:,15:25].mean(axis=-1)
tone_idxs = trial_starts + 0.5
stim_idxs = trial_starts + 2
resp_idxs = trial_starts + 4
for x1,x2,x3,y1,y2,y3 in zip(tone_idxs,stim_idxs,resp_idxs,tone,stim,resp):
ax[4].plot((x1,x2,x3),(y1,y2,y3),color='k',marker='o',
label = "Collapsed data" if x1==tone_idxs[0] else "")
for a in ax[:-1]: a.set_xticklabels([])
for a in ax: a.set_xlim(ax[0].get_xlim()); a.legend(loc='upper right')
for a,char in zip(ax,"ABCDEFG"):
a.set_title("$\\bf{(" + f"{char}" +")}$", loc="right")
ax[2].set_ylabel("Fluorescence ($\Delta$F/F0 units)")
ax[-1].set_xlabel("Time (s)")
fig.show()