Supplementary Figure 1. Non-phosphorylated A and B adaptors are ligated to the ends of phosphorylated, polished, double-stranded genomic DNA fragments. The A and B adaptors differ in both nucleotide sequence and the presence of a 5’ biotin tag on the B adaptor. Nicks are present at the 3’-junctions of each of the adaptors and the library fragment are filled in by the strand-displacement activity of Bst DNA polymerase. Streptavidin- biotin interactions are used to remove fragments flanked by homozygous adaptor sets (A/A and B/B) and to generate single stranded library templates. Fragments are bound to Streptavidin beads; unbound material (composed of homozygous A/A adaptor sets, which lack biotin) is washed away. The immobilized fragments are then denatured; both strands of the B/B fragments remain immobilized through the biotinylated B adaptor, while A/B fragments are washed free and used in subsequent sequencing steps. Replicate library preparations were observed to yield coverage of the genome and oversample with CV’s of 5% or less.
Supplementary Figure 2. Size distribution of nebulized DNA sample. Sharp flanking peaks are upper and lower reference markers.
Supplementary Figure 3. Kineticmodeling of single well. Assumption: 10 million DNA copies per bead, [DNA] = 0.3 μM.
Supplementary Figure 4. Chemical cross-talk modeling. At t=0, [DNA]well 1 = 0.3 μM, [DNA]well 2= 0.
Supplementary Figure 5. Detailed error rates in sequencing a mixture of 6 test fragments, as a function of homopolymer length. Single base error rates are referenced to the total number of single bases sequenced. For each homopolymer,the error rateis referenced to the total number of bases sequenced that belong to homopolymers of that length.
Supplementary Figure 6. Typical histogram of signal intensities for negative and positive flows.
Supplementary Figure 7. Average of the flow signals ascribed to various homopolymers for the mapped reads of the M. genitalium run discussed in the paper.
Supplementary Figure 8. Detailed error rates in sequencing an M. genitalium library, as a function of homopolymer length. As for test fragments, single base error rates are referenced to the total number of single bases sequenced; for homopolymers,the error rateis referenced to the total number of bases sequenced that belong to homopolymers of each length. The error rates are shown for individual reads and after the consensus sequence was formed using all reads, without Z-score restriction.
Supplementary Figure 9. Depth of coverage as a function of genome position for the M. genitalium run. Slightly lower coverage in isolated regions is due to the presence of repeat regions excluded in the mapping.
Supplementary Figure 10. Correlation between predicted and observed quality scores for a sequencing run of C. jejuni (data not shown).
Supplementary Figure 11. Read lengths of paired end reads. Note this was for a 21 cycle run so the average length is commensurate with the lower number of cycles.
Page 1 of 12Manuscript 2005-05-05204JMRothberg Supplementary Figures.doc