Computational Drug Discovery and Design

visualization, and machine learning. The code and data used in the

following section is freely available athttps://github.com/psa-lab/

predicting-activity-by-machine-learning.

2 Materials

2.1 Python
Interpreter

To perform the analyses described in section3, a recent Python

[23] version (3.5 or newer) is required (Python 3.6 is recom-

mended). A Python installer for all major operating systems (Mac

OS, Windows, and Linux) can be downloaded fromhttps://www.

python.org/downloads/.

2.2 Python Libraries
for Scientific
Computing

The following list specifies the Python libraries used in this chapter,

the recommended version number, and a short description of their

use:

l NumPy version 1.13.0 or newer (http://www.numpy.org);

numerical array library [24].

l SciPy version 0.19.0 or newer (https://www.scipy.org);

advanced functions for scientific computing [25].

l Pandas version 0.20.1 or newer (http://pandas.pydata.org);

handling of CSV files and working with data frames [26].

l Matplotlib version 2.0.2 or newer (https://matplotlib.org); 2D

plotting [27].

l Scikit-learn version 0.18.1 or newer (http://scikit-learn.org/

stable/); algorithms for machine learning [28].

l MLxtend version 0.7.0 or newer (http://rasbt.github.io/

mlxtend/); sequential feature selection algorithms [18].

The scientific computing libraries listed above can be installed

using Python’s in-built Pip module (https://pypi.python.org/

pypi/pip) by executing the following line of code directly from a

Mac OS/Unix, Linux, or Windows MS-DOS terminal command

line:

pip install numpy scipy pandas matplotlib scikit-learn

pydotplus mlxtend

If you encounter problems with version incompatibilities, you

can specify the package versions explicitly, as shown in the following

terminal command example:

pip install numpy==1.13.0 scipy==0.19.0 pandas==0.20.1

matplotlib==2.0.2 scikit-learn==0.18.1 pydotplus==2.0.2

mlxtend==0.7.0

314 Sebastian Raschka et al.