Optimization of Rflp Based Sensors

Restriction fragment polymorphism (rflp) is the basis for a wide range of analytical tools in biology. In this paper we propose maximum entropy as an appropriate metric for the analysis of terminal-rflp techniques for bacterial based on the 16S small rDNA subunit, with the following specific results:

1. The computation of a maximum entropy a posteriori probability density is a convex optimization, where information gain establishes an upper bound on sensor performance.

2. For a given apriori probability density function, each candidate restriction enzyme can be assigned a performance index based on Kullback-Leiber divergence.

3. Optimization of t-rflp utilizing multiple restriction enzymes is formulated and solved as a convex, geometric programming problems.

4. Limits to the robust performance of t-rflp sensors subject to measurement error are established.