if anyone is interested in the algorithms used (Score 5, Interesting)

We published a series of papers evaluating correlated mutation algorithms, including SCA, which is the algorithm used in this pair of Nature papers. I haven't had a chance to look closely at the two new papers, but we found that SCA performed rather poorly when compared to other algorithms that calculate covariance from a multiple sequence alignment. SCA has a troubling tendency to assign high scores to pairs of columns of a multiple sequence alignment that have random sequence in them.

PDFs of our papers, and Java code implementing 4 different correlated mutation algorithms including SCA, are at my web site:

http://www.afodor.net

The references are:

Anthony A. Fodor, Richard W. Aldrich. On Evolutionary Conservation of Thermodynamic Coupling in Proteins. JBC 279(18):19046-19050, 2004

John P. Dekker, Anthony Fodor, Richard Aldrich and Gary Yellen. A pertubation-based method for calculating explicit likelihood of evolutionary co-variance in multiple sequence alignments. Bioinformatics 20:1565-1572, 2004

Anthony A. Fodor and Richard W. Aldrich. Influence of Conservation on Calculations of Amino Acid Covariance in Multiple Sequence Alignments. Proteins 56(2): 211-221, 2004

The last paper contains a comparison between SCA and three other correlated mutation algorithms.

As I said, I haven't had a chance to look carefully or critically at the new papers. (It takes me a LONG time to read a paper critically :-> This Slashdot thread will be likely long archived before I finish thinking about these papers!). But this particular algorithm aside, people who are interested in bioinformatics and contact prediction may find the math behind the correlated mutation algorithms interesting.

Anthony

Email: anthony.fodor(remove this and put in an at symbol)gmail.com
http://www.afodor.net/