Hidden Markov Model Essay

Introduction: The human genome contains a wealth of information about our bodies. We know that there are genes which lead to long, healthy lives. We also know that there are genes which can lead to short and painful ones, and everything in between. The difficulty, however, lies in finding which genes in which states lead to which health outcomes. Genome sequencing is becoming faster, cheaper, and more accessible. This means that we are increasing our store of information about the human genome at astounding rates. And science needs to catch up. This store of data contains enormous amounts of useful information which may save millions of lives. But first we need to unlock it. This is where data mining can be incredibly useful. Using mining techniques scientists can search for similar gene sequences across species, find where our DNA diverged from our ancestors, and even gain clues into the creation of new and extremely powerful medications for genetic diseases. One particularly powerful tool in this data mining effort are hidden Markov models, a form of data mining most useful for gaining information from time series data. It has extensive applications within bioinformatics, including protein folding, DNA classification, and the alignment of bio-sequences. In essence, a hidden Markov model which is given the outcomes of a causal chain, can
In an HMM, we have a series of states, and a corresponding series of emissions, occurring over time. Because it is a hidden model, we cannot directly observe the states, but we can observe the by-products of these states, the emissions. Through training this model, we will be able to begin to gain information about the states based solely on the observations. It is this ability which makes HMMs such a powerful tool in many applications of data