Virtual Cell Virtual Cell WebsiteSchool of Health Sciences WebsitePurdue University Website
About the Virtual CellDownload VCExamplesTheoryUser ManualHelpExecute VC Simulation
Introduction
How to generate synthetic datasets with random noise


What is random noise?

The term "noise" originated in communication engineering. Any random or orderly disturbance that is of significantly different characters (frequency and amplitude) to the main signal is called "noise". Larger the signal-to-noise ratio (S/N or SNR), the better the situation is. Addition of "noise" to an ordinary mathematical function will distort the smoothness of the function. There are subdivisions of the term denoting its character such as "white-noise", "pink-noise etc.  

Random noise consists of a large number of transient disturbances with a statistically random time distribution. These transient disturbances have oscillating magnitudes as a function of time, according to a normal (Gaussian) curve. Also called Gaussian noise and random Gaussian noise.

Biological data and random noise

Data obtained from most physical systems, i.e., engineering systems, biological systems (think of the air velocity in a wind tunnel or the number of colonies formed in a seeded petri-dish) will show "scatter", that is, the data do not conform to a pure mathematical functional form. For the average of a large number of point estimates however, the data may closely approximate a functional form. Unless a suitable random disturbance model (such as a random noise generator) is included in a mathematical model for an engineering or a biological system will not predict  this "scatter" one would expect from actual measurements. Noise of appropriate magnitude can be added to make the data generated by a mathematical model look more "natural" by embedding the nature of unpredictability.

Generation of synthetic data by VC  

VC can be used to generate synthetic biological data by adding random noise to the model predicted responses (the -n command line option must be used to add noise to the predicted responses). Parameters RERR, the relative error can be used to predetermine the relative error involved. Parameter RNGS, the random number generator seed can be used to yield different synthetic data sets for the same set of input parameters stipulating the nature of biological uncertainty involved in repeated experiments.


NOISE:  Specifies the magnitude and origin of noise to be added to generate the synthetic data.


Generating synthetic biological data  (Tutorial #16)


Tip: Follow the use of keywords and observe the trends in the output variation.
School of Health Sciences
Purdue University
Disclaimer

Last updated: 10 June, 2011
Webmaster