Basic Usages

Introduction

pheatmap is a Python package for ploting heatmap with annotation bars. It just like pheatmap package of R.

Installation

You can install pheatmap by pip:

pip install pheatmap

Run the above command, pip will intall pheatmap and its dependencies, numpy, pandas and matplotlib.

Quick Start

Before create heatmap, we need to import some essential packages.

import pandas as pd
import numpy as np
from pheatmap import pheatmap

Simple Heatmap

First, we can create a DataFrame, each row means a gene, and columns are samples.

mat = pd.DataFrame(
    np.random.randint(0, 10, 600).reshape(30, 20),
    index=[f"G{i}" for i in np.arange(30)],
    columns=[f"sample{i}" for i in np.arange(20)]
)
# Show some expression values
mat.iloc[:, :5].head()

	sample0	sample1	sample2	sample3	sample4	sample5
G0	4	9	4	6	6	1
G1	1	4	6	3	2	1
G2	8	8	9	1	0	8
G3	2	5	9	3	2	3
G4	4	6	8	3	2	1

Then, we can plot a simple heatmap for these genes’ expression on samples:

fig = pheatmap(mat, colnames_style={"rotation": 90})

sample_heatmap

Annotation Heatmap

Annotation columns

If we have some information for samples, such as experiment v.s. control or sample source, then we can and these information to the plot by annotation_col parameter.

# Create DataFrame, store samples' information
anno_col = pd.DataFrame({
    "source": np.repeat(["source1", "source2", "source3", "source4"], 5),
    "group": np.repeat(["experiment", "control"], 10)
})
fig = pheatmap(mat, annotation_col=anno_col, colnames_style={"rotation": 90}, legend_bar_space=1.5)

col_anno_heatmap

We can point colors for every annotation bar by annotation_col_cmaps parameter. annotation_col_cmaps need a dict, the keys of dict are the anno_col’s columns, and the values of dict are the colors. If the number of colors are not enough for the categories’, pheatmap will cycle the colors to fullfill the number of categories.

fig = pheatmap(
    mat, annotation_col=anno_col,
    annotation_col_cmaps={"group": ["#7FC97F", "#FB9A99"]},
    colnames_style={"rotation": 90}, legend_bar_space=1.5
)

col_anno_heatmap2

Annotation Rows

We can also add annotations for genes. For example, we want to check the degree of DNA methylation for genes. We can put the degree of DNA methylation on the left of heatmap. By the way, we have the information of clusters for genes and want to see it at the same time.

anno_row = pd.DataFrame({
    "DNA-methyl": np.linspace(0, 1, 30),
    "gene_clusters": ["CNS"[i % 3] for i in np.random.randint(0, 10, 30)]
})
fig = pheatmap(
    mat, 
    annotation_col=anno_col,
    annotation_col_cmaps={"group": ["#7FC97F", "#FB9A99"]},
    annotation_row=anno_row,
    annotation_row_cmaps={"DNA-methyl": "Purples", "gene_clusters": "Set2"},
    colnames_style={"rotation": 90}, 
    legend_bar_space=1.5, 
    wspace=0.15
)
fig.savefig("row_col_anno_heatmap.svg")

row_col_anno_heatmap