Advertisement
Research Article

Gene Expression in the Rodent Brain is Associated with Its Regional Connectivity

  • Lior Wolf mail,

    wolf@cs.tau.ac.il (LW); ruppin@post.tau.ac.il (ER)

    Affiliation: Blavatnik School of Computer Science, Tel-Aviv University, Tel-Aviv, Israel

    X
  • Chen Goldberg equal contributor,

    equal contributor Contributed equally to this work with: Chen Goldberg, Nathan Manor

    Affiliation: Blavatnik School of Computer Science, Tel-Aviv University, Tel-Aviv, Israel

    X
  • Nathan Manor equal contributor,

    equal contributor Contributed equally to this work with: Chen Goldberg, Nathan Manor

    Affiliation: Blavatnik School of Computer Science, Tel-Aviv University, Tel-Aviv, Israel

    X
  • Roded Sharan,

    Affiliation: Blavatnik School of Computer Science, Tel-Aviv University, Tel-Aviv, Israel

    X
  • Eytan Ruppin mail

    wolf@cs.tau.ac.il (LW); ruppin@post.tau.ac.il (ER)

    Affiliations: Blavatnik School of Computer Science, Tel-Aviv University, Tel-Aviv, Israel, School of Medicine, Tel-Aviv University, Tel-Aviv, Israel

    X
  • Published: May 05, 2011
  • DOI: 10.1371/journal.pcbi.1002040

Reader Comments (1)

Post a new comment on this article

This paper used a incorrect approach

Posted by JunweiWang on 11 May 2011 at 14:13 GMT

Whether the correlation of correlation can suggest there exists association or connectivity between structures, please consider it twice? Please ask for help from statisticians.

No competing interests declared.

RE: This paper used a incorrect approach

liorwolf replied to JunweiWang on 11 May 2011 at 18:22 GMT

Thanks JunweiWant for your comment. I am not sure I fully understand it. Please try to explain your concern in more details.

Also, much of the statistical techniques we use follows the footsteps of previous work. In your view, are we using it incorrectly or is the approach incorrect in general.

Competing interests declared: (author)

RE: RE: This paper used a incorrect approach

PLoS_CompBiol replied to liorwolf on 02 Jun 2011 at 10:07 GMT

The journal has received the following response from JunweiWang:
Thank you for replying! I am sure this is not the correct way. Pearson correlation measure the linear association. Then the authors calculated the correlation based on this pearson correlation. I have studied the reference the authors provided. That paper could be published because they do many biological experiments. The paper in PLoS computational Biology calculate the results, the method of correlation have used incorrectly. Firstly, one is that the assumption of pearson correlation is normal distribution. According to my test of microarray data, the distribution is very complex. Maybe it is Gamma distribution and so on. The other assumption is the observations is independent, but correlation among genes could not be independent, as the gene network are known. Secondly, the structure correlation is calculated based on gene expression' correlation could not illustrate the connectivity between structures, but it could not indicate the connectively.
The correct way should be pairwise correlation between gene expression of different structure could be illustrated this connectivity.

No competing interests declared.

RE: RE: RE: This paper used a incorrect approach

PLoS_CompBiol replied to PLoS_CompBiol on 09 Jun 2011 at 15:27 GMT

The journal has received the following response from the author, Lior Wolf:

Thank you for giving much thought to our work.

With regards to the correlation method, I am not sure I understand all your arguments and I did not understand your suggestion.

Some of your arguments are based on assertions I don't necessarily agree with. For example, Pearson correlation can be used in non Gaussian situations. The p-value might not be accurate in such cases and this is why we also provide empirical p-values.

If you would like to learn more about the correlation test, please refer to:
Kaufman et al, 2006
French L, Pavlidis, 2011

No competing interests declared.