Transcript Sequence Registry

In many computational genomic studies, including large-scale expression profiling, it is useful to have a comprehensive collection of transcript sequences. The Reference Sequence [NCBI, 1999] project at NCBI is an effort to identify a non-redundant set of full-length mRNA sequences contained within GenBank [Benson et al., 1999] entries. Currently, RefSeq consists of approximately 6,300 human mRNA sequences and approximately 2,000 mouse mRNA sequences. Since each RefSeq entry requires human review in order to meet desired quality standards, the rate of introduction of new RefSeq sequences is limited. At the same time, high-throughput genomic sequencing efforts, along with highthroughput annotation, are increasing the rate of introduction of (putative) new genes, full-length and partial, into GenBank. As such, it is unlikely that RefSeq will provide a comprehensive collection of transcript sequence from GenBank for some time. In order to provide a comprehensive repository of transcript sequences to support our internal projects, we have developed a system, the Transcript Sequence Registry (TSR), which automatically excises transcript sequences from both RefSeq and GenBank, and then stores the sequences, along with associated information, in a relational database. The Transcript Sequence Registry is intended to be the foundation of a gene-oriented resource; though unlike UniGene [Wheeler et al., 2000], expressed sequence tags (EST) are excluded.