Identification of rat genes by TWINSCAN gene prediction, RT-PCR, and direct sequencing

Abstract

The publication of a draft sequence of a third mammalian genome--that of the rat--suggests a need to rethink genome annotation. New mammalian sequences will not receive the kind of labor-intensive annotation efforts that are currently being devoted to human. In this paper, we demonstrate an alternative approach: reverse transcription-polymerase chain reaction (RT-PCR) and direct sequencing based on dual-genome de novo predictions from TWINSCAN. We tested 444 TWINSCAN-predicted rat genes that showed significant homology to known human genes implicated in disease but that were partially or completely missed by methods based on protein-to-genome mapping. Using primers in exons flanking a single predicted intron, we were able to verify the existence of 59% of these predicted genes. We then attempted to amplify the complete predicted open reading frames of 136 genes that were verified in the single-intron experiment. Spliced sequences were amplified in 46 cases (34%). We conclude that this procedure for elucidating gene structures with native cDNA sequences is cost-effective and will become even more so as it is further optimized.

Figure 1

A TWINSCAN prediction (green, subsequently identified as rat aspartylglucosaminidase) in which sequences from the single intron (purple), terminal exon (red), and UTR (blue) primer pairs all yielded spliced alignments. Primers are shown as tall pink blocks at the same level as the sequence they yielded (the left single-intron primer did not yield high-quality sequence). TWINSCAN makes use of the blocks of alignment from the human genome (black) and the mismatches and gaps within the blocks (red) in predicting the most likely gene structure.