Building a de novo Chicken Genome Annotation with PacBio Transcriptome Sequencing

Identification of the transcriptome forms the foundation of building a complete genome annotation. RNAseq experiments often depend on having accurate predictions of transcripts to quantify expression. However, the vast majority of species have incomplete transcriptome annotations making it difficult for researchers to extract biologically relevant information from their data. The ability to rapidly build genome annotations with accurate transcriptome predictions has become increasingly important due to the accelerated production of genome assemblies for previously unsequenced organisms.

Various strategies have been employed to identify transcripts, however many of them suffer from the inability to accurately predict whole transcript structures. Short read RNAseq offers high throughput and high coverage, which makes it suitable for differential expression estimates and gene discovery. However, short read data produces ambiguous predictions of alternative splice variants.

We used Pacific Bioscience transcriptome sequencing to generate reads spanning whole transcripts. Using this data we have been able to predict alternative splice variants and anti-sense genes with high confidence. We have built a pipeline that combines long and short read data to create a de novo genome annotation for Chickens. Our de novo annotation provides information on transcript location, structure and tissue type expression. It contains thousands of predicted transcripts and expressed regions that are not present in publicly available annotations.

Our annotation will be available on the National Avian Research Facility website (http://www.narf.ac.uk).  

This work was funded by the BBSRC, Wellcome Trust and University of Edinburgh.