Denoiser is designed to address issues of noise in pyrosequencing data. Denoiser is a heuristic variant of PyroNoise. Developers of denoiser report a good agreement with PyroNoise on several test datasets.

Denoiser uses three clustering steps:

1. cluster perfect reads on the sequence level and get an initial abundance distribution of the common OTU
2. cluster near-perfect reads to the common OTUs using flowgram alignments and a strict clustering threshold
3. cluster bad reads not previously clustered using a relaxed threshold

To start up the denoiser environment on Bio-Linux, you need to type:

If you wish to use the sffinfo command referred to in the documentation, you will need to install gsAssembler from Roche on your machine.

A different version of the denoiser.py script is run from within the qiime environment than from within the denoiser environment. To run denoiser from within the qiime environment, exit from the denoiser environment if you are in it, and then type qiime on the command line.