def plot_pca2(df, filters=None):
# index has sample names, colors+markers should be constructed with filter
if filters is None:
filters = create_diagonal_trues(df)
markers, colors = construct_point(filters.shape)
else:
markers, colors = construct_point(filters.index)
fig, ax = plt.subplots(figsize=(10, 8))
def plot_kmeans(df,
k,
n_clusts,
title,
col_names=('Principal Component 1', 'Principal Component 2')):
"""
Args:
df:
k:
col_names:
Returns:
"""
# markers = ['o', 'v', 's', 'd', 'P', '+', '8', '*', "X", 'x', "D", 'd', 7, 6,
# "$:)$", "$:($", "$=$", "$?$",
# "$!$"]
markers, colors = helpers.construct_point(n_clusts)
# colors = [colors]
for n in range(n_clusts):
k_f = k == n, 0
k_t = k == n, 1
pc1 = df.loc[k_t[0], col_names[0]]
pc2 = df.loc[k_t[0], col_names[1]]
plt.scatter(
pc1,
pc2,
# c=[colors[n]], marker=markers[n],
edgecolor='black',
label='Cluster {}'.format(n))
this_clust = df[k_t[0]]
print(this_clust.iloc[:, :2], '\n')
plt.legend()
label_count = 0
indices = pc1.index.to_list()
for x, y in zip(pc1, pc2):
plt.annotate(
indices[label_count], # this is the text
(x, y), # this is the point to label
textcoords="offset points", # how to position the text
xytext=(0, 10), # distance from text to points (x,y)
ha='center',
size=6) # horizontal alignment can be left, right or center
label_count += 1
plt.savefig(
'/Users/coltongarelli/Desktop/laura\'s LPP stuff/kmeans_{}_plot.pdf'
.format(title))
def plot_pca(df, rows, batch: bool = False):
"""Plots PCA data on a scatter plot where x = PC1 and y = PC2
Plot the first two principal components of PC analyzed data on a scattermap. Plotting PCs can be used as a measure
for batch effect or correction. PCA can also be used to determine correlation of a list of interest genes.
Notes:
- Perform PCA on specific genes to utilize PCA for non-batch analysis
Args:
df: a DataFrame containing PC analyzed data to plot on a scatter plot (x = PC1, y= PC2)
rows: A dictionary that maps each sample or batch to each PC data experiment/sample in df.
batch: consider changing to 'group' for generic grouping
Returns:
Figure and Axes objects
"""
markers, colors = construct_point(df.shape[0])
cdict = dict(zip(rows.keys(), colors))
marker = dict(zip(rows.keys(), markers))
fig, ax = plt.subplots(figsize=(10, 8))
if batch:
# consider changing batch to 'group' for generic grouping
labels = ['Batch {}'.format(i + 1) for i in range(len(colors))]
else:
labels = df.index.to_list()
for k, v in rows.items():
ax.scatter(x=df.loc[v, 'Principal Component 1'],
y=df.loc[v, 'Principal Component 2'],
c=[cdict.get(k)],
marker=marker.get(k),
label=labels[k],
s=55)
plt.xlabel("Principal Component 1", fontsize=15)
plt.ylabel("Principal Component 2", fontsize=15)
for i in df.index.to_list():
ax.annotate(i, (df.loc[i, ['Principal Component 1']] + .05,
df.loc[i, ['Principal Component 2']] + .05))
# ax.legend(loc='upper right', bbox_to_anchor=(1.04, 1))
plt.tight_layout()
ax.legend().remove()
return fig, ax
def plotly_3d_pca(df, batch: bool = False):
markers, colors = construct_point(df.shape[0])
# cdict = dict(zip(rows.keys(), colors))
# marker = dict(zip(rows.keys(), markers))
# if batch:
# # consider changing batch to 'group' for generic grouping
# labels = ['Batch {}'.format(i+1) for i in range(len(colors))]
# else:
# labels = df.index.to_list()
x = df.T.loc['Principal Component 1']
y = df.T.loc['Principal Component 2']
z = df.T.loc['Principal Component 3']
fig = plotly.offline.plot(
go.Figure(go.Scatter3d(x=x, y=y, z=z, mode='markers', text=labels)))
fig.update_traces(textposition='top center')
fig.update_layout(title_text='3D PCA of LPP Nanostring by groups')
fig.update_layout(scene=dict(xaxis_title='Principal Component 1',
yaxis_title='Principal Component 2',
zaxis_title='Principal Component 3'))
return fig