Python in neuroscience

Computation is becoming essential across all sciences, for data acquisition and analysis, automation, and hypothesis testing via modeling and simulation. As a consequence, software development is becoming a critical scientific activity. Training of scientists in programming, software development, and computational thinking (Wilson, 2006), choice of tools, community-building and interoperability are all issues that should be addressed, if we wish to accelerate scientific progress while maintaining standards of correctness and reproducibility. The Python programming language in particular has seen a surge in popularity across the sciences, for reasons which include its readability, modularity, and large standard library. The use of Python as a scientific programming language began to increase with the development of numerical libraries for optimized operations on large arrays in the late 1990s, in which an important development was the merging of the competing Numeric and Numarray packages in 2006 to form NumPy (Oliphant, 2007). As Python and NumPy have gained traction in a given scientific domain, we have seen the emergence of domain-specific ecosystems of open-source Python software developed by scientists. It became clear to us in 2007 that we were on the cusp of an emerging Python in neuroscience ecosystem, particularly in computational neuroscience and neuroimaging, but also in electrophysiological data analysis and in psychophysics. Two major strengths of Python are its modularity and ability to easily “glue” together different programming languages, which together facilitate the interaction of modular components and their composition into larger systems. This focus on reusable components, which has proven its value in commercial and open-source software development (Brooks, 1987), is, we contend, essential for scientific computing in neuroscience, if we are to cope with the increasingly large amounts of data being produced in experimental labs, and if we wish to understand and model the brain in all its complexity. We therefore felt that it was timely and important to raise awareness of the emerging Python in Neuroscience software ecosystem amongst researchers developing Python-based tools, but also in the larger neuroscience community. Our goals were several-fold: