29th Nov 2010

The Arabidopsis Information Resource (TAIR) is pleased to announce the release of the latest version of the Arabidopsis genome annotation (TAIR10). The latest release builds upon the gene structures of the previous TAIR9 release using RNA-seq and proteomics datasets as well as manual updates informed by cross species alignments, peptides and community input regarding missing and incorrectly annotated genes.

The TAIR10 release contains 27,416 protein coding genes, 4827 pseudogenes or transposable element genes and 1359 ncRNAs (33,597 genes in all, 41,666 gene models). A total of 126 new loci and 2099 new gene models were added.

Eighteen percent (5885) of Arabidopsis genes now have annotated splice variants. Updates were made to 1184 gene models of which 707 had CDS updates. There were 41 gene splits and 37 gene merges.

Gene annotation utilized available Proteomics data (Baerenfaller et al., 2008 and Castellana et al., 2008) and RNA-seq data from the Ecker and Mockler labs (Lister et al. 2008, Filichkin et al. 2010). RNA-seq data was mapped to the Arabidopsis genome using Tophat, hashmap or supersplat. After quality and low complexity filtering a total of ~200 Million RNA-seq reads were successfully mapped to the genome of these ~9 million represent spliced reads. Proteomics data and spliced RNA-seq reads were provided to Augustus and the resulting gene models categorised and manually reviewed. Validated gene updates, novel genes and novel splice variants were incorporated into the TAIR10 release. Additional spliced RNA-seq reads not already incorporated into gene models by Augustus were supplied to TAU. The resulting TAU models were again reviewed for potential novel splice variants. Transcript assemblies were generated via Cufflinks using all spliced reads and unspliced reads from the Ecker sets. Transcript assemblies were filtered and compared to existing gene models resulting in the addition of 56 novel genes. Additional new Proteome data provided to us by Katja Baerenfaller was used to directly update 24 gene models.

Gene models created using the Gnomon pipeline were provided to TAIR by NCBI. Reanalysis of these models for TAIR10 resulted in the addition of 11 novel, 67 alternative splice variants and 164 updates to existing genes.

We would like to thank all those who contributed to the latest release by providing submissions for new and incorrectly annotated genes.

TAIR wishes to thank Cornell University for use of the computer clusters at the Cornell Center for Advanced Computing (CAC).