Collecting genomic, transcriptomic, epigenomic and interaction data across genomes and taxa leads to new hypotheses, suggests explanations for age-old phenomena and answers questions we did not even know existed. Four articles in this issue review the contribution of large-scale analyses to our understanding of genome evolution. Crucially, they also counter the view that 'high-throughput' science essentially equates to cranking a handle. Large-scale analyses rely on the same iterative process as traditional science, which involves an interesting question, a hunch, the appropriate technology and a critical appraisal of the results.

In their Review on p487, Koonin and Wolf assess the complex constraints that operate at different levels of organization — from the nucleotide sequence to the phenome. With so many species and variables to consider, good judgement is key to correctly interpreting patterns in the data. Addressing other evolutionary questions requires a careful selection of genomes to study. Genome evolution is influenced by interactions with other species, notably so in bacteria–host associations. As Toft and Andersson explain on p465, genomics has propelled the field of microbial evolution to a new level of understanding, but this is largely due to the conscious decision to switch from the study of free-living microbes to selected host-associated species. Yeast genomics is making comparable strides: Dujon's Review (p512) reveals how sequences from a range of phylogenetically targeted species are providing information about eukaryotic evolutionary genomics.

Finally, as Hawkins, Hon and Ren discuss in a Review (p476), valuable biological insights will emerge by integrating 'omics' data. Processing large volumes of data demands, more than ever, scientific input, as the question being asked dictates which data and analysis methods are used.