Python matplotlib: all about fonts

Fonts output into pdf as text, not shape, to be recognized in Illustrator:

import matplotlib
matplotlib.rcParams['pdf.fonttype'] = 42
matplotlib.rcParams['ps.fonttype'] = 42

Change fonts style for all, change the default:

import matplotlib as mpl
matplotlib.rcParams[''] = 'Arial'

Change fonts style for labels, tick marks, and titles: (for each plot, not changing the default)

plt.xlabel('XXXX', fontsize=12, fontname='Arial')
plt.ylabel('XXXXX', fontsize=12, fontname='Arial')


plt.title('XXXXX', fontsize=14, fontname='Arial')

Change legend font style: (for each plot, not changing the default)

plt.legend(prop=matplotlib.font_manager.FontProperties(family='Arial', size=12, weight='bold', style='normal'))


Liftover bam files

The most straightforward way is using CrossMap.

Taking from hg19 to hg38 as example:

pip install CrossMap bam -a hg19ToHg38.over.chain input.bam output
#.bam extension will be added automatically

genome liftover chain files can be downloaded here: (change according to your needs)

It is suggested to always use ‘-a’ option according to the CrossMap website.


Python Jupyter notebook share the variable across notebooks

yourVar = 'data or your variable'
%store yourVar
del yourVar # only deletes the variable in this notebook but not in store

In the second notebook:

%store -r yourVar # if you have a variable with the same name, it will rewrite it.

To list and delete the stored variable:

%store #list variable in store
%store -d yourVar #delete the variable in store


Unix check if a file exists

test -f yourFile

if [ -f "$FILE" ]; then
    echo "$FILE exists."
    echo "$FILE does not exist."

If need to match wildcard in the file names

if test -n "$(find /dir/to/search -maxdepth 1 -name 'files*' -print -quit)"
    echo found
    echo not found


Ignore row with only NaN in plotHeatmap – deepTools

If there are NaN in the output from computeMatrix, the generated heatmap is not sorted and a warning message stating Mean of empty slice will show up.

To overcome this, those null values need to be replaced using 0 in the computeMatrix step by --missingDataAsZero tag.

computeMatrix scale-regions -S -R xxx.bed --missingDataAsZero -m xxx -b xxx -a xxx --numberOfProcessors xx -o xxx.gz

plotHeatmap -m xxx.gz -out xxx.png