Taxonomic assignment using read sets trimmed to different lengths.

Results for six set lengths between 170 (the minimum acceptable read length considered in this study) and 350 nucleotides are compared with the set of 400 nucleotide reads; thus the cells of the table always represent a comparison with the characteristics of the 400 nucleotide set. ‘Identical hits’ shows the number and percentage of reads getting identical hits irrespective of read length; ‘Identical hits; no hits removed’ shows the same calculations but omitting any read that did not get a hit in either of the compared sets; ‘Richness’ shows the total number of VT and the richness as a percentage of that in the 400 nucleotide set; ‘Mean sample-wise similarity’ was a comparison of the community composition of the same sample in the 400 nucleotide set and in the shorter trimmed sets. This was calculated using two metrics: Bray-Curtis dissimilarity and Pearson’s correlation coefficient. Samples containing fewer than 10 reads were removed from the data set, leaving 42 samples and 4 508 reads. VT abundance was normalised (converted to proportions) prior to similarity calculations.