解碼基因為人類帶來了期望，因為可藉此了解遺傳及環境引起的人類進化、疾病和DNA內部的構造功能。人類基因組序列宣示將在2005年完成基因定序，為加速如此龐大的基因鹼基序列定序尋找基因，需仰賴電腦科技運算的基因預測程式，因此許多程式陸續地被開發。然而，基因預測工具相當多，目前這些程式都無法相當正確的尋找基因，還未能完全真正解決許多基因結構預測的問題。 本研究結合統計特徵的計算分析與候選exon區段計分演算兩種方式進行基因尋找。此方法的主要目的是定義exon序列及基因序列邊界，而其邊界是藉由比較基因序列區段的鹼基組成統計特徵，以及邊界的函數進行劃分。 本研究之基因預測的程式整合了電腦運算及統計技術，可在配合實驗室研究下幫助生物學家分析新定序的序列，並致力克服人類基因組複雜的結構，迅速搜尋基因，能以統計量化數據窺得序列全面的特徵，不需複雜的統計模式進行推估，仰賴訊號正確性，未來希望調查exon中鹼基出現頻率規則使得基因預測在實驗研究過程中化繁為簡，達成生物學家生物實驗研究先機，對於醫學研究影響深遠。; Decoding gene bringing an expectation to understand how the heredity and environment affect human evolution, human disease and DNA structure and function. Human Genome Project publicizes that entire human gene sequencing will be finished in 2005; however the progress is uncertain so that many gene prediction programs are designed continuously to speed up gene finding in massive amount of genomic sequences. Although gene prediction programs are developed for the complex structure of human genomic sequences, but a accuracy program still be needed. The research combined statistical characteristic analysis and exon region score algorithm to find gene region. The purpose of our method is to define exon region and gene region boundary. The boundary is generated by comparing base composition of gene region and scoring function of boundary. The program integrates computer science and statistical analysis which tries to solve the problem of complex human genome structure could help biologists to find gene rapidly, the sequences are numeral to analysis global structure of genomic sequences without complex statistical model to depend on signal to estimate. Future work is hopefully to help biologists to investigate exon frequency and discover exon composition rule to predict gene automatically.
