Machine learning in life science

Machine learning in life science

Machine learning’s popularity is increasing among bioinformaticians and biologists as it gives interesting results and has become more accessible than ever. A machine learning model can now be easily applied on a given dataset using R or Python packages. For example, the Python package Scikit-learn provides several algorithms (Random Forest, Support Vector Machine – SVM -, regression model and much more) and good documentation.

Even deep machine learning (neural networks with multiple layers or convolutional networks for example) is more accessible nowadays. Several deep learning frameworks exist to facilitate the usage of those more complex models: some are low-level such as Theano and TensorFlow while others are high-level such Lasagne, Blocks, Mariana and Keras. Lasagne, Blocks and Mariana are actually built on top of Theano. Wider lists of available frameworks can be found on Wikipedia or on deeplearning.net website.

The Mariana framework was in fact developed by one of IRIC’s student and tested by members of the bioinformatic platform. Check the video below to learn more about this framework designed to be an “Extendable Python Machine Learning Framework build on top of Theano that focuses on ease of use.”

I hope that the next time you’ll be tempted by machine learning, you’ll dive into it with enthusiasm even if you’re not an expert, knowing that there exist some options especially for you out there.



Video1. Tariq Daouda presenting Mariana

By | 2016-11-08T09:30:05+00:00 May 18, 2016|Categories: Machine learning|0 Comments

About the Author:

I’ve started in biochemistry but it is as a bioinformatician that I’ve been having fun for several years now : whether doing data analysis and visualization in R, building interactive web interfaces in javascript or exploring machine learning in python.

Leave A Comment