Poster: Bioinformatics

Abs # 911: EST data mining and expression pattern clustering

Presenter:	Jung, Woosuk , jungw@konkuk.ac.kr
Authors	Jung, Woosuk  (A)   Kim, Sun  (B)
Affiliations:	(A): Konkuk University (B): Indiana University

The recent development of high-through put sequencing technology has provided a new tool to investigate genomes at the DNA level and the transcription level. More than 1,600,000 ESTs have been identified from the cDNA libraries of major crops (soybean, rice, wheat, barley and maize) and model plants (Arabidopsis and Medicago truncatula) (http://www.ncbi.nlm.nih.gov/dbEST/dbEST_summary.html). Since we assume that there are known sequences of the family in other organisms, we may apply a straightforward scheme that performs search with the set of known sequences against the EST databases and reports matches with some cutoff values. However, there are two nontrivial problems with this straightforward scheme: 1. The search extensiveness problem: since ESTs are short, relatively high-error sequences, some true matches may be short and low similarity. If we lower the database search criteria like E value or bit score, then higher false positives will be included in the search result. 2. The base call distinction problem: since we are searching for sequences that belong to the same family, it is necessary to rely on base differences in very short regions for identification of distinct gene sequences. Here, we present a novel computational method for mining distinct genes of a certain family in the EST databases. The overview of the data mining procedure will be explained with an example of the MYB transcription factor gene family. Recently MYB transcription factor gene family in Arabidopsis has been extensively studied and more than 90 members are known. We searched rice MYB genes with the MYB gene sequences in Arabidopsis and the expression frequencies of rice MYB genes will be discussed.