Matched filter for signature extraction in bioinformatics data set

In many bioinformatics research and applications, we need to extract interesting signals or to compare between signals in large multivariate data sets. In most cases those signals are buried in “noise”. Finding methods to increase signal-to-noise ratio is an important statistical problem studied for many years and there exist a set of well developed methods for noise reduction and signal boosting. In large scale multivariate bioinformatics data sets, when the signature of the signal is known, at least approximately, the concept of a matched filter can be applied to increase the signal-to-noise ratio dramatically. This, in turn, makes the automatic signal extraction and detection possible and reliable. Further more, matched filter is the optimal linear filter with the present of white noise.