The SILVA database v.106 and ARB software v. 5.2 were used to identify and test specificity of the different primers in silico[30] (link), [31] (link), [51] (link). An internally developed AmpliconGenerator v.0.1 (http://sf.net/projects/amplicongenerator) was used to characterize nucleotide variability along the reference sequence. The software was also used to generate in silico amplicons and to assess the conservation of biodiversity as a function of sequence similarity. IDT SciTools OligoAnalyzer 3.0 (Integrated DNA Technologies, Coralville, IA) was used to estimate melting temperatures of the primers as well as the presence of hairpins, self- and hetero-dimers in the candidate primers. Sequences from the environmental samples were processed through the bioinformatic pipeline AmpliconNoise according to Quince and co-workers [6] . As a first step in the pipeline, low quality reads were removed, defined as shorter than 200 nt or having inadequate signal intensity. Subsequently, noise from the flowgram and PCR-generated errors were removed using established probabilistic iterative algorithms [6] , [52] (link) and chimeras removed using the program Perseus [6] . The reads were then clustered together using the complete linkage-clustering algorithm implemented in FCluster, based on pairwise distances as calculated by NDist [6] , and aligned to the SILVA and NCBI databases using BLAST for the purpose of taxonomic annotation. The same software package was also applied to build operational taxonomic units (OTUs). Since the choice of the pipeline can influence the results of metagenetic studies and because there is no consensus on which pipeline to use, the results produced with AmpliconNoise were compared to results produced by an alternative commonly used pipeline, QIIME. In QIIME, original reads were filtered based on the quality score, but no further denoising procedure was applied. Further, in order to remove potential chimeras, OTUs representing singletons were discarded.
Free full text: Click here