Insights from Genomics into Bacterial Pathogen Populations
Sample collection. A biological sample (e.g., blood) is collected. Culture. Bacterial colonies are isolated from the sample by culturing on appropriate media. DNA Preparation. DNA is extracted from the colonies and a DNA library is prepared ready for sequencing. High-Throughput Sequencing. Millions of short sequence reads are yielded, typically several hundred nucleotides long or less. To reconstruct the genome, one of two approaches is generally adopted. Mapping to Reference Genome. In reference-based mapping, the short sequences are mapped (i.e., aligned) to a reference genome using an algorithm (e.g., , ). Preferably the reference genome is high quality, complete, and closely related. The pie chart illustrates that not all reads necessarily map to the reference genome (e.g., because of novel regions not present in the reference). Filtering. Short reads cannot be mapped reliably to repetitive regions of the reference genome, so these are identified and filtered out. Sites that are problematic for other reasons (e.g., because too few reads have mapped or because the consensus nucleotide is ambiguous) are also filtered out. The pie chart illustrates that some portion of the reference genome does not get called due to filtering. In the mapped genome, these positions will receive an ambiguity code (i.e., N rather than A, C, G, or T). De novo Assembly of Contigs. An alternative to mapping is de novo assembly, in which no reference genome is used. An algorithm (e.g., , ) is used to assemble short reads into longer sequences known as contigs. The number and length of contigs will depend on general factors such as the length of sequence reads and the total amount of DNA sequence produced, as well as local factors such as the presence of repetitive regions. The pie chart shows an example of the proportion of all reads that assemble into contigs of a given length. Alignment. For further analysis, it is necessary to align local regions (e.g., genes) or whole genomes using appropriate algorithms (e.g., –). There is a trade-off in computational terms between the length of region and the number of sequences that can be aligned. Sequence Analysis. The two approaches produce sequence alignments that represent pairwise alignments against a reference (mapping) or multiple alignments one to another (de novo assembly). These alignments can be analyzed directly, or processed further to detect variants such as single nucleotide polymorphisms, insertions, and deletions. The pie charts are meant to be illustrative only, and were produced from data in .