Bovine Microarray Consortium
Abstract presented at ISAG 2006, Porto Seguro, Brazil August 20-25, 2006
A Bovine Whole Genome Long Oligonucleotide Expression Array
CHRISTINE G. ELSIK1, ERIC ANTONIOU2, SCOTT C. FAHRENKRUG3, JAMES M. REECY4, RUSSELL D. WOLFINGER5, JEREMY F. TAYLOR2
1Texas A&M University, College Station, USA. 2University of Missouri, Columbia, USA. 3University of Minnesota, St. Paul, USA. 4Iowa State University, Ames, USA. 5SAS Institute Inc., Cary, USA.
Contacting author's email: c-elsik@tamu.edu
The design and synthesis of 24,000 long oligonucleotide probes comprising the first generation bovine whole genome expression array has been completed. The array includes 16,846 probes designed from ESTs that were aligned to homologous vertebrate proteins and to the 6X bovine genome assembly (BGA). The protein alignment step allowed removal of chimeric ESTs and selection of ESTs that match protein coding genes. After grouping ESTs with homologous proteins, each group was assembled using the TIGR pipeline, TGICL, which includes pre-clustering (MEGABLAST) and assembly steps (CAP3). Alignment of the resulting contigs and singletons to the BGA indicated that repetitive elements were abundant among the protein-coding EST clusters, despite repeat removal using RepBase. To remove protein-coding transposable elements without inadvertently removing transcription factors, the contigs and singletons were searched for PFAM domains found in transposable elements, excluding zinc-finger domains. Following transposable element filtering, 36,547 clusters (22,740 contigs; 13,807 singletons) could be aligned to the BGA which grouped the clusters to form 16,849 unique gene loci. Probes were designed at Illumina from 16,846 ESTs derived from clusters aligned to unique genes.
The following criteria were used in EST selection and probe design:1) predicted constitutive exon, 2) polymorphism avoidance, 3) minimal distance to 3' end of protein coding region, and 4) optimal Tm and specificity. Constitutive exons were predicted by identifying the regions of each gene's sequence with the highest EST coverage. To sample as many different genes as possible our goal was to represent one constitutive exon rather than multiple alternative exons per gene. The probe set was supplemented with oligos designed from 703 predicted RefSeq genes, 5943 reproductive tissue ESTs with a BGA but no protein alignment, and 504 +/- controls. Probes were annotated using descriptions of homologous proteins, including Gene Ontology. A link to probe sequences and annotations will be available here in the near future.