Issues of EST in whole genome gene discovery

[cyberpostdoc]

EST is short for expressed sequence tags, where partially sequenced cDNA fragments are used to represent gene expression. It has the power of gene discovery on a genomic level, However, there are limitations of this technology. Of course the advantage of EST is that it provides direct evidence of the expression of a gene. But there are several drawbacks I could think of:

1. mRNA abundance limitation: it is estimated that genes have an average express ion level below 5% of the mRNA population is less likely to be detected by EST technology. Therefore, genes that are expressed at extreme low level won't be able to be detected by EST, so the 27K in this sense should be an lower estimate (if we only focus on this matter). Now, one of the other technique is more sensitive, it is called SAGE, for serial analysis of gene expression, where tags at 3' end of genes are collected and sequenced to represent mRNA expression.

2. development stage limitation: some of the genes in our genome may only express at certain development stage, therefore, if EST libraries cannot cover this development stage, both for technical and ethical reasons, there is no way these gene can be detected.

3. diseased stage limitation: on the same line of #2, but from an opposite point of view, some mRNA might be a result of diseased stage, not neccissary be a normally expressed gene. Or aberrent expression that are not constitute normal behavior of human genome, hence, enriched EST libraries from cancer cell lines (for example) will likely to result in over estimation of human transcriptome.

4. sequencing error: this could post extreme difficulties for bioinformatics to
further map these ESTs back to genome to identify genes. Especially from paralog genes.

Finally, besides gene discovery, EST can be used to study post transcriptional control on a genomic level, which have been shown very successful, including studies in alternative splicing and alternative polyadenylation. Pioneer of these include Michael Zhang at CSHL, Chris Burge at MIT, and Chris Lee at UC.

Please add.