Abstract
▼

Splice site identication in pre-mRNA is an important task for identifying new genes and their structures and functions. Gene identication also plays an important role in molecular biology for diagnosing disease and drug discovery. For this reason, much research has been performed to predict genes st...

Splice site identication in pre-mRNA is an important task for identifying new genes and their structures and functions. Gene identication also plays an important role in molecular biology for diagnosing disease and drug discovery. For this reason, much research has been performed to predict genes structure and function. In this thesis we propose a splice site prediction method using the RSCU (relative synonymous codon usage) values of introns and exons. Our method encodes DNA sequences with the RSCU values of 59 codons (leaving out three stop codons and two unique codons) to search for introns and exons. We then embed 118 dimensional feature vectors into the input vectors of binary SVM (support vector machine) to establish a learning model and classify test data. Furthermore, we extend our encoding approaches on nucleotide's chemical properties in the era of evolutionary relationship analysis on mtDNA sequence of 8 different species. Extensive performance study shows that our method can provide better performance than existing encoding methods based on several performance criteria such as classication accuracy, sensitivity (Sn), specicity (Sp), auROC (area under Receiver Operating Characteristics) and degree of similarity dissimilarity. Key words: RSCU, SVM, gene classication, splice site, HS3D, mtDNA, Hexag- onal Model i

참고문헌 (0)

1. 이 논문의 참고문헌 없음

이 논문을 인용한 문헌 (0)

1. 이 논문을 인용한 문헌 없음