Statistical mechanics of transcription-factor binding site discovery using hidden Markov models (Q540552)

The paper can be considered as being part of a series of articles dealing with the use of results from statistical mechanics to the study of different problems in bioinformatics. In the present work one investigates the use of hidden Markov models (HMM) to find the minimum amount of data needed to train an algorithm to learn the binding sites of a transcription factor. After presenting the connection between HMMs and the statistical mechanics of hard rods, maximum likelihood equations via virial expansions as well as an analytic expression for calculating the Fisher Information are derived. This expression is then used to formulate a criteria for how much samples are needed to learn the binding sites of a transcription factor of a given specificity.