Bioinformatics challenges of new sequencing technology

Abstract

New DNA sequencing technologies can sequence up to one billion bases in a single day at low cost, putting large-scale sequencing within the reach of many scientists. Many researchers are forging ahead with projects to sequence a range of species using the new technologies. However, these new technologies produce read lengths as short as 35-40 nucleotides, posing challenges for genome assembly and annotation. Here we review the challenges and describe some of the bioinformatics systems that are being proposed to solve them. We specifically address issues arising from using these technologies in assembly projects, both de novo and for resequencing purposes, as well as efforts to improve genome annotation in the fragmented assemblies produced by short read lengths.

Figure 1

Effect of repeats on long and short read assemblies. The middle section of the figure represents a set of repeats longer than 30 bp within a 50-kbp region of the Yersinia pestis CO92 genome. The blue (red) boxes represent repeats longer (shorter) than 800 bp. An assembly of this region using Sanger data would result in the set of contigs represented at the top, correctly resolving the short repeats and only breaking at the boundaries of long repeats. Furthermore, paired-ends (indicated by thin lines connecting the contigs) would provide long range connectivity across repeats. The assembly generated from short read sequencing data (bottom line) is considerably more fragmented, breaking at all repeat boundaries, and lacking long range connectivity caused by the absence of paired-end data.

Figure 2

Repeats longer than 30 bp in the genomes of Bacillus anthracis Ames and Yersinia pestis CO92. Tic marks on the outer concentric circles correspond to direct (same strand) and palindromic (opposite strand) repeats; every tic mark represents a breakpoint (where a gap would occur) for an assembly based on reads of 30 bp or less. Contigs generated from short read sequencing data would cover 97% and 93% of these genomes, respectively, with N50 sizes of 30 387 and 25 894 bp. The fractions of these genomes covered by unique segments longer than 10 kbp (1 kbp) are 84% (96%) and 66% (91%), respectively.

Figure 3

Fragmentation of a gene caused by fragmented genome assembly. This example shows how a five-exon gene (blue), known from cDNA sequencing or a closely related species, maps to three different short contigs (red). The assembly fails to capture all of exons e2 and e5, which run off the ends of contigs. Exon e5 maps in reverse orientation to contig c2. The cDNA would allow us to order the contigs as c1-c3-c2r (where ‘r’ means reversed), but without a full-length cDNA, we would have no indication of how these contigs were related to one another.