I feel like a knowledge sponge since I came back to Paris last summer. In few months, I followed most of the MOOCs on genetics or systems biology. On Coursera, two were especially helpful: Introduction to Genetics and Evolution, and Network Analysis in Systems Biology. Another big change was to fully switched from Matlab to Python. The simplicity of this programming language even gave me the motivation to start free programming courses with HackYourPhD: the Open Geek sessions!
Anyway, one of my new projects at the Institut Pasteur focuses on Network Based Stratification, a machine learning technique combining Non-Negative Matrix Factorization (NNMF) with networks of Protein-Protein Interaction (PPI). A first version was introduced by Hofree and colleagues (Nat. Meth. 2013) to classify different types of cancer tumors. Since our goal is to stratify large cohort of patients with Autism Spectrum Disorders in more homogeneous subgroups, I am currently adapting mathematically their approach for the case of non-somatic mutation. The tool is in Python—it’s called StratiPy ;)—so that it could be available in open source and cheap to run on many computers (Matlab licences are quite expensive those days…). You can have a look on my online iPython Notebook or download the code on Github. All comments are welcome!
Leave a Reply