Benchmarking Next-Generation Transcriptome Sequencing for Functional and Evolutionary Genomics

• Published in: Molecular biology and evolution 2009-12, Vol.26 (12), p.2731-2744
• Main Authors: Gibbons, John G., Janson, Eric M., Hittinger, Chris Todd, Johnston, Mark, Abbot, Patrick, Rokas, Antonis
• Format: Article
• Language: English
• Subjects: Aedes - genetics; Aedes aegypti; Alternative Splicing - genetics; Amino Acid Substitution - genetics; Animals; Anopheles - genetics; Anopheles gambiae; Aquatic insects; Comparative analysis; Contig Mapping; Evolution; Evolution, Molecular; Female; Gene Expression Profiling - methods; Genes; Genomics; Genomics - methods; INDEL Mutation - genetics; Insects; Male; Molecular biology; Mosquitoes; Oligonucleotide Array Sequence Analysis; Protein synthesis; Proteins; Proteome - genetics; Reference Standards; RNA, Messenger - genetics; RNA, Messenger - metabolism; Sequence Analysis, DNA - methods; Sequence Homology, Nucleic Acid
• ISSN: 0737-4038; 1537-1719
• DOI: 10.1093/molbev/msp188

Next-generation sequencing has opened the door to genomic analysis of nonmodel organisms. Technologies generating long-sequence reads (200-400 bp) are increasingly used in evolutionary studies of nonmodel organisms, but the short-sequence reads (30-50 bp) that can be produced at lower cost are thought to be of limited utility for de novo sequencing applications. Here, we tested this assumption by short-read sequencing the transcriptomes of the tropical disease vectors Aedes aegypti and Anopheles gambiae, for which complete genome sequences are available. Comparison of our results to the reference genomes allowed us to accurately evaluate the quantity, quality, and functional and evolutionary information content of our "test" data. We produced more than 0.7 billion nucleotides of sequenced data per species that assembled into more than 21,000 test contigs larger than 100 bp per species and covered ∼27% of the Aedes reference transcriptome. Remarkably, the substitution error rate in the test contigs was ∼0.25% per site, with very few indels or assembly errors. Test contigs of both species were enriched for genes involved in energy production and protein synthesis and underrepresented in genes involved in transcription and differentiation. Ortholog prediction using the test contigs was accurate across hundreds of millions of years of evolution. Our results demonstrate the considerable utility of short-read transcriptome sequencing for genomic studies of nonmodel organisms and suggest an approach for assessing the information content of next-generation data for evolutionary studies.