Bioinformatics: Python vs Other Languages
Many of us think Python is great. In data-rich scientific fields (including Biology), high-level languages like Python and R remain widely used both for method development and data analysis. On numeric data, array programming in Python is fast – thanks to NumPy. However, operations on non-numeric data were slow – the reason why many of the widely used Bioinformatics tools were written in low-level languages.
What is not so great about Bioinformatic tools written in low-level languages
Don’t get me wrong. Any programming language as a tool that serves the purpose and needs is useful. I’ve been a user of many popular bioinformatics tools written in low-level languages and I’ve only great things to say about a large majority of them. They serve their intended purpose quite efficiently. But when I needed to make a small change in the inner workings of a tool to suit my new needs in specific cases, it was difficult to make the desired changes efficiently. This can be true for an average computational biologist whose primary languages of choice are high-level languages.
Fast and easy operations on non-numeric biological data in Python using BioNumpy
My colleagues from Norway developed BioNumpy, which loads non-numeric biological data into NumPy arrays enabling fast computations. On a range of common bioinformatic tasks, the nightly runs of benchmarks show how BioNumpy is faster than many of our favourite Bioinformatic tools. This isn't just a minor improvement; it's a potential game-changer. The point is not that our favourite tools should be replaced – but it shows that BioNumpy can serve as a foundational layer for the development of new bioinformatics tools (with non-numeric array computation needs), where BioNumPy lowers the overhead and entry barrier for developers. By significantly decreasing the execution time of common bioinformatics tasks in Python, BioNumPy can at the same time benefit all the practitioners in the field (who use Python primarily for their data analysis), offering a substantial boost in efficiency. In addition to the many good examples on the BioNumPy documentation, I showed in this Colab notebook one example of how I used BioNumpy to perform end-to-end analysis on biological sequence data from exploratory analyses to machine learning. Give BioNumpy a try today – and see the difference for yourself.
