Too big to fail. Or not so big.

In the Nature issue of August 2013, two very renowned authors, Barzel & Barabasi reported a method that claims to reconstruct network topologies [1]. Their paper, titled Network link predictions by global silencing of indirect correlations describes a method that takes into account direct and indirect correlations between nodes and tries to silence the indirect effects by exploiting some properties of dynamical correlations in the network. As they write in the abstract “The method receives as input the observed correlations between node pairs and uses a matrix transformation to turn the correlation matrix into a highly discriminative silenced matrix, which enhances only the terms associated with direct causal links.”

In a note to the Editors [2] it has been reported that the aforementioned method not only is a variant of a previously published one, namely modular response analysis (MRA). But also its implementation is erroneous. Barzel&Barabasi use a statistical similarity measure to take into account of global network responses to perturbations. That happens to lead to overestimated performance.

I am particularly sensitive to such a topic for two reasons.
The first one is that I have been working on topology reconstruction from data in my paper published on PLOS [3].
The second is that such a reconstruction problem is an open problem especially in biology. Therefore any claim gets under the spotlight of scientists and reviewers quite fast.

The note to the editors which can be read in [2] is, to some extent a relief, that sheds light on the subject and clarifies the wrong claim from the authors (who are supposed to be rock stars in the topic). They probably forgot to mention that they were silencing indirect correlations and relevant literature and errors in implementation.
Unexpected at all.

Reference

[1] Network link predictions by global silencing of indirect correlations
[2] bastiaens_et_al_kholodenko_about_barabasi_nature_biotech2015
[3] Discovering main genetic interactions with LABnet Lasso Based network inference

About these ads

About Piggy

I am Piggy and I spend my life reading about math and of course eating. I love science and I support my flatmate who provides me problems to solve and, well, food.
This entry was posted in General and tagged , , , , , , , , . Bookmark the permalink.

Leave a Reply

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

WordPress.com Logo

You are commenting using your WordPress.com 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 )

Google+ photo

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

Connecting to %s