Information Content of Individual Genetic Sequences

@article{Schneider.Ri,
author = "T. D. Schneider",
title = "Information Content of Individual Genetic Sequences",
journal = "J. Theor. Biol.",
volume = "189",
number = "4",
pages = "427-441",
note = "\htmladdnormallink
{http://www.ccrnp.ncifcrf.gov/\~{}toms/paper/ri/}
{http://www.ccrnp.ncifcrf.gov/\~{}toms/paper/ri/}",
year = "1997"}

US patent 5867402
Material in this paper is covered by US patent 5867402. Programs using these methods may not be redistributed or used without a signed agreement with the National Institutes of Health. Please contact us at: http://www.ccrnp.ncifcrf.gov/~toms/contacts.html

This method of analyzing binding sites can be distinguished from other methods by the following criteria.

Consensus sequences can be immediately rejected, see the paper Consensus Sequence Zen.
A variety of ad hoc methods are non-additive, these can be immediately rejected. Shannon chose his function to be addititve and it is the only one that has this property.
Berg and von Hippel's (Stormo's) method does not give results in bits. If you flip a coin, according to this method, you could get thousands of 'bits' of information. Despite the claims, it is not information theory and it does not properly connect to thermodynamics because it confuses non-specific binding states with specific binding states. It also ignores the inequality in the Second Law of Thermodynamics.
Starting from Berg and von Hippel, if one sets the genomic frequencies to equiprobabile one gets a method that is linearly proportional to Ri. These can be rejected as not having a natural zero coordinate. The zero coordinate corresponds to the Second Law of Thermodynamics. This method, as with the original Berg and von Hippel method, therefore must use an arbitrary cutoff.
Neural network training methods assume that places where we do not know anything are not sites. This has been demonstrated to be wrong. An example is the missed Fis sites in the tgt/sec promoter.

Having set all these criteria, there is no other method.

Additional discussion is in Measuring Molecular Information.

Also, see the paper

@article{Erill.ONeill2009,
author = "I. Erill
 and M. C. O'Neill",
title = "{A reexamination of information theory-based methods for
DNA-binding site identification}",
journal = "BMC Bioinformatics",
volume = "10",
pages = "57",
pmid = "19210776",
year = "2009"}

For more infomation see: Individual Information Theory and Sequence Walkers

color bar Small icon for Theory of Molecular Machines: physics,
chemistry, biology, molecular biology, evolutionary theory,
genetic engineering, sequence logos, information theory,
electrical engineering, thermodynamics, statistical
mechanics, hypersphere packing, gumball machines, Maxwell's
Daemon, limits of computers

Schneider Lab
origin: 1997 December 23
updated: 2010 Mar 31
color bar