【備忘録】PythonのBokehでBandのグラフを描く【可視化】

pip install bokeh

from bokeh.models import Band, ColumnDataSource

from bokeh.plotting import figure, show

from bokeh.io import output_notebook, show

from bokeh.models import ColumnDataSource, HoverTool, LinearColorMapper

from bokeh.plotting import figure

from bokeh.models import SaveTool

def generate_dataset(mu):

    # 正規分布の乱数生成

    rng = np.random.default_rng()

    # x軸の値

    x = list(range(100))

    # mu: 各点における母平均

    # 各点における母標準偏差

    sigma = [abs(rng.normal(1, 1)) for i in range(100)]

    # 母平均と母標準偏差を用いて各点における1000個のデータセットを生成

    data = [rng.normal(m, s, 1000) for m, s in zip(mu, sigma)]

    #1000個のデータセットから平均と標準偏差を求める

    mean = [np.mean(d) for d in data]

    std = [np.std(d) for d in data]

    # プロット用に標準偏差を用いてバンドの上限と下限を計算

    upper = [mean[i] + std[i] / 5 for i in range(len(mean))]

    lower = [mean[i] - std[i] / 5 for i in range(len(std))]

    # Return

    return x, mean, upper, lowe

mu = [rng.normal(0.03 * i, 0.025) for i in range(100)]

x, m, upper, lower = generate_dataset(mu)

# bokeh graph

p = figure(width=900, height=350, tools=["save"])

p.title.text = "Change in expression of Gene X in response to pseudotime"

p.xgrid.grid_line_alpha = 0

p.ygrid.grid_line_alpha = 0

# axis label

p.xaxis.axis_label = 'Pseudotime'

p.yaxis.axis_label = 'TPM'

# series 1

mu = [rng.normal(0.03 * i, 0.025) for i in range(50)] + [rng.normal(1.5 - 0.002 * i, 0.025) for i in range(50)]

x, m, upper, lower = generate_dataset(mu)

source = ColumnDataSource(data=dict(x=x, m=m, lower=lower, upper=upper))

p.line("x", "m", color="green", line_width=2, source=source, legend_label="Control")

band = Band(base="x", lower="lower", upper="upper", source=source, fill_alpha=0.1, fill_color="green", line_color="green")

p.add_layout(band)

# series 1 max/min

max_1 = max(upper)

min_1 = min(lower)

# series 2

mu = [rng.normal(0.008 * i, 0.025) for i in range(50)] + [rng.normal(0.4 - 0.008 * i, 0.025) for i in range(50)]

x, m, upper, lower = generate_dataset(mu)

source = ColumnDataSource(data=dict(x=x, m=m, lower=lower, upper=upper))

p.line("x", "m", color="red", line_width=2, source=source, legend_label="Sample")

band = Band(base="x", lower="lower", upper="upper", source=source, fill_alpha=0.1, fill_color="red", line_color="red")

p.add_layout(band)

# series 2 max/min

max_2 = max(upper)

min_2 = min(lower)

# x/y axis range

y_max = max(max_1, max_2)

y_min = min(min_1, min_2)

p.x_range=Range1d(0, 100)

p.y_range=Range1d(y_min - 0.1*abs(y_min), y_max + 0.1*abs(y_max))

# legend position

p.add_layout(p.legend[0], "right")

output_notebook()

show(p)