Multiscale denoising of biological data: A comparative analysis
Academic Article
Overview
Research
Identity
Other
View All
Overview
abstract
With the advancements in computing and sensing technologies, large amounts of data are collected from various biological systems. These data are a rich source of information about the biological systems they represent. For example, time-series metabolic data can be used to construct dynamic genetic regulatory network models, which can be used not only to better understand the interactions among different genes inside a cell, but also to design intervention strategies that can cure or manage major diseases. Also, copy number data can be used to determine the locations and extent of aberrations in chromosome sequences which are associated with many diseases such as cancer. Unfortunately, measured biological data are usually contaminated with errors that mask the important features in the data. Therefore, noisy biological measurements need to be filtered to enhance their usefulness in practice. Conventional linear low-pass filtering techniques are widely used because they are computationally efficient and can be implemented online. However, they are not effective because they operate on a single scale, meaning that they define a specific frequency, above which all features are considered noise. Real biological data possesses multiscale characteristics, i.e., may contain important features having high frequencies (such as sharp changes) or noise occurring at low frequencies (such as correlated or colored noise). Filtering such data requires multiscale filtering techniques that can account for the multiscale nature of the data. In this work, different batch as well as online multiscale filtering techniques are used to denoise biological data. These techniques include standard multiscale (SMS) filtering, online multiscale (OLMS) filtering, translation invariant (TI) filtering, and boundary corrected TI (BCTI) filtering. The performances of these techniques are demonstrated and compared to those of some conventional low-pass filters (such as the mean filter and the exponentially weighted moving average filter) using two case studies. The first case study uses simulated dynamic metabolic data, while the second case study uses real copy number data. Simulation results show that significant improvement can be achieved using multiscale filtering over conventional filtering techniques.
