The IceCube neutrino telescope is currently constructed in the deep ice near the geographic South Pole. At completion IceCube will consist of more than 4500 photomultiplier tubes which record the Cherenkov light from secondary charged particles produced in neutrino nucleon interactions, in order to detect high energy neutrinos coming from cosmological objects. The recorded signals are used to reconstruct the energy and direction of the incident neutrino. The core of the custom made software components, ranging from simulation to data compression and likelihood reconstruction is written in C++. The program flow structures are exposed to Python using Boost.Python. This allows to control the data processing flow in Python which offers a great flexibility. Data analyses can be performed using Python and libraries like numpy, scipy, matplotlib, pytables and pyROOT. For this purpose the main data structures are exposed to Python, as well. The talk will give an overview about the control flow of the IceCube software chain, driven by Python, and how reconstructed data is analysed using Python and extension libraries.