Figure 1

Reproducibility of 3' tag digital gene expression profiling. 1a - Pearson correlation coefficient matrix of all 8 HBRR libraries prepared at two Mayo locations (Minnesota and Florida), sequenced in 35 lanes of 5 different runs on two generations of Genome Analyzer (I and II). The rows and columns are all 35 lanes in 5 HBRR sequencing runs, and named using the corresponding library names (L1-L8). The same libraries in each run have been grouped together for visual benefits. The correlation coefficient was calculated using Log2 transformed tag counts. The tag counts of zero were coded as missing data. The actual numbers of the correlation coefficient are listed in the Additional File 2. The color in each squire of the matrix reflects the pair-wise lane-to-lane degree of correlation of gene expression levels; 1b - Concordance of gene detection. Gene expression levels are represented by the raw number of reads (dark red) and number of reads per million (yellow). More than 70% of the genes were repeatedly detected in all 35 lanes. 1c - The relationship between gene detection and expression levels. Genes that were detected in less than 35 lanes were lower expressed at levels of 1-2 CPMT, or 0.35-0.7 copies per cell; 1d - The histogram of gene expression levels.