English: a) Contig Length Distribution. Histograms of the contig lengths illustrate the number of contigs within a certain size fraction for assemblies of Illumina reads with quality filtering (red), Sanger sequenced reads (yellow) and reads from pyrosequencing (blue). Only contigs greater than 500 bp are shown, the x-axis is log scale. Assemblies were generated for different community complexities (10 genomes, 100 genomes, 400 genomes). b) Overall Accuracy of the Contigs. The overall accuracy of the contigs is summarized using different measures of chimericity. Bars to the left illustrate the percentage of all of the contigs that are chimeric, bars in the middle show the percentage of all of the contigs that have a Contig Score less than 95%, and to the right contigs that have a Contig Score less than 99%. Contig Score represents the percent identity between the contig and the derived reference genome. Contigs from Illumina reads are red, contigs from Sanger reads are yellow and contigs from pyrosequencing are blue. In general there was a slightly higher proportion of Illumina contigs that were chimeric, however they had higher contig scores. c) Contig Accuracy across Contig Lengths. These combined strip plots show the degree of chimericity (upper plot) and contig score (lower plot) for each contig in the assemblies, each dot represents one contig. They are grouped into size bins. The degree of chimericity is the proportion of reads in a contig that are derived from the ‘wrong’ genome and thus make the contig chimeric. Contig Score represents the percent identity between the contig and the derived reference genome. Again contigs from Illumina assemblies are in red, from Sanger assemblies are in yellow and from pyrosequencing assemblies are in blue. For all sequencing technologies and communities, longer contigs are more accurate.