How I met neuroinformatics


Previously working at the Brain & Spine institute (or Institut du Cerveau et de la Moelle épinière i.e. ICM in french), I was exposed to neuroinformatics. This discipline is an excellent example of what means data science today.

What is neuroinformatics ?

Basically, neuroinformatics could be defined as bioinformatics applied to neurology. The reason to create a special term for it is that neurology has some very specific questions and issues.

A good example of neuroinformatics project is the Blue Brain project in Switzerland. The team achieves to reconstruct with very fine details a small region of the brain. Thanks to their software, they can simulate brain stimuli in this region almost perfectly. The project is now expending to other regions of the brain. I had the opportunity to discuss with Pr Sean Hill, one of the co-director of the project, who explained us that after 13 years they are still looking for the right tools and organization to manage this meta-project efficiently…

Is neuroinformatics “just” another application of bioinformatics ?

When I heard the first time about neuroinformatics, my first though was it was just bioinformatics applied to brain data, as it could be applied to heart, skin or any specialized organ/cells. I soon discovered I was wrong.

First, neuroinformatics use imaging and stimulation of neurons/region of the brain, in addition to genomic and molecular biology data. Processing these data uses specific processes of acquisition, cleaning and analysis, especially imaging data, that could be very heavy.

On top of that, processing all these data (and probably others) to solve biological questions on brain required a specific knowledge 1) of the data 2) of the question. Their methods are not completely new, but this specific knowledge is requiered to apply them in neuroinformatics questions. In that way, neuroinformatics in an embodiment of a data science. You can not apply it just-the-same-algorithm than any other topics. It required a special training that even someone who is trained in biology and use similar methods in other topics, could not jump into neuroinformatics easily.

Ok, but what do they do ?

Questions in neuroinformatics are often around how the brain works, or does not woks in case of diseases.

Some researches are around understanding a healthy brain, how usual informations (images, sound, social interactions…) are processed and how it leads us to decision. This kind of researches could go into very tricky question such as the level of “strategy” of individuals. For instance, I heard a very interesting talk about Theory of Mind where people were looking at performances of players in a game of “hide and seek” with only to possible locations to hide.  The idea was to evaluate the level of I-know-that-you-know-that-I-know of players against other players or against computer programs in different scenario. The data and the methods are highly variable from one subject to another. The outcome are many-fold, from just a better of understanding of what we are, to developing new methods to educates people, making them more efficient… or manipulate them. Of Course, it is also helpful to understand diseases.

Other kind of research, focuses on possible treatement to diseases such as Parkinson disease or Alzheimer disease. These research often focuses on comparing heathy brain to patient brain to find differences. Here, imagery data are often used, combined with genomic and even functional networks to discriminate patients from healthy control. Machine learning approaches are more often classification of clustering than prediction. The outcome is, of course to cure these diseases. In could be a new therapeutic target, a way to repair brain damage, or to compensate them or detecting people at risk to develop these diseases and develop preventive treatments.

As in many clinical projects, patient recruitement is a huge issue due to ethical concerns and having a good cohort of patients and controls is a project by itself, beside designing experiments and finding a way to analyze the data.

To conclude this article, here is an example of software developed by neuroinformatics: Clinica is a software that allow to acquire different type of data (imaging, genetics…) from various subjects and to make specific analysis. It shows the complexness of the data and the wide range of skill to process them, which makes neuroinformatics a real data science, even as a sub-field of bioinformatics.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s