I would like to reproduce the statistical result in this paper: http://www.ncbi.nlm.nih.gov/pubmed/24947363

They said "the M-statistic test in ProtoArray Prospector was applied to the normalized intensities as reported previously. For each probe, M test 1) produced the number of samples (M-statistic counts) with the intensity greater than or equal to the M-statistic threshold in two groups and 2) computed the P value (the significance of the difference between two groups) from the sample sizes and the M-statistic counts in the two groups".

But there is NO clear description or reference for the mentioned M statistic. I searched for a while and found one relevant literature: Love, B., in Predki,P.F.(Ed.), The Analysis of Protein Arrays in Functional Protein Microarrays in Drug Discovery. 2007, USA: CRC Press

Anyway so far I have no access to this chapter. My question number 1 is: Is there anyone here have good openly accessible reference for this M statistics (formula)? If so, please let me know.

Back to my original aim/issue: The data is published in GEO with accession number is GSE50866. The software is free (I do not know if it has limited features in free version) and the manual is also available in the company website. I could not put the link here since I do not have enough reputation to do that.

I downloaded ProtoArray Prospector and looked at the manual but could not find the right place to get the formula for M statistic, so I could not re-produce the result with this software. Any tip is highly appreciated.


A paper from Mark Gerstein's lab describes how they implemented the M-statistic in analyzing ProtoArrays:

  1. For each protein, compare each healthy signal to each disease signal. The number of disease signals which are above the highest healthy signal equals M*.
  2. A p-value is determined by computing the probability of having M greater than M*, using the hypergeometric distribution.
  3. Then the number of disease signals which are above the second highest healthy value is counted, giving a new M* and a new p-value.
  4. Keep going, comparing to the next highest healthy signal, until you've reached the lowest healthy signal.
  5. Use the lowest p-value obtained this way as the significance of this protein.

Here's the link to their paper for more details:


It's not open-access, unfortunately, so you'll need a subscription to Journal of Proteome Research to read it. Here is an ungated link.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.