A study for determination of deletion position under dependent copy number data

• Main Authors: Chun-ShengLin, 林駿昇
• Other Authors: Mi-Chia Ma
• Format: Others
• Language: en_US
• Published: 2016
• Online Access: http://ndltd.ncl.edu.tw/handle/6bpg44

碩士 === 國立成功大學 === 統計學系 === 104 === Non-invasive Prenatal Testing (NIPT) is a relatively safe way for pregnant women to test whether fetuses have congenital disease, because small segments of fetuses’ DNA will enter into blood of pregnant women through by umbilical cord and placenta. Doctors extract the blood of pregnant women to obtain fetuses’ DNA, in order to detect whether fetuses have a congenital disease. The purpose of this study is to find a statistical method which can detect DNA gene deletion on chromosome 22, as the detection method of DNA deletion disease in medicine. Olshen and Venkatraman (2004) proposed Circular Binary Segmentation (CBS) method connecting an ordered gene in a ring to find out the position with significant differences in average of reads count as change points. This study consider correlated data of reads count between genes, mixed ideas of Hotelling's T-squared test, extended CBS to multi-dimension cases. In addition, we will use statistical simulation methods and real examples to compare Type I error rate, power and Type III error rate of both methods.