@MastersThesis{Son01, author = {S{\"o}ren Sonnenburg}, title = {Hidden Markov Model for Genome Analysis}, school = {Humboldt University}, year = {2001}, type = {Project Report}, ps = {http://sonnenburgs.de/soeren/publications/Son01.ps.gz}, pdf = {http://sonnenburgs.de/soeren/publications/Son01.pdf.gz}, abstract = { A Hidden Markov Model (HMM) is a Markov chain, where each state generates an observation. One only sees the observations, and the goal is to infer the \emph{hidden} state sequence. HMMs have been applied successfully e.g.~to Speech Recognition tasks and to biological sequence analysis. They have recently been used for gene finding and for identifying homologous sequences. The aim of our work was to create a reliable, flexible and efficient tool that can be used for \emph{genome analysis}. Our implementation exhibits some particularly interesting features, such as Viterbi and Baum-Welch training algorithms, customizable model structures and support for higher order Hidden Markov models. It has been particularly tuned for large numbers of observations. To speed-up the computations it also supports multi-processor systems. By this work we have created the basis for developing more sophisticated algorithms for genome analysis in subsequent projects. } }