Description
- Abstract:
- The five-year survival rate in breast cancer patients drops from 99% at Stage I to about 23% at Stage IV. Early detection of diseases such as breast cancer can help to prolong the life expectancy of patients. One method for disease prediction is the identification of pathogenic variations. Studies have suggested that the pathogenic variants are enriched in the conserved region of the gene. The first theme of my work is the proposal of a novel evolutionary measure, called the variation number, to define evolution conservation. The approach utilizes multiple sequence alignment from orthologous sequences as well as phylogenetic information among species. Statistical analysis suggests that the mean variation number is smaller in human pathogenic mutations than the neutral ones in BRCA1. However, the variation number itself may not completely separate pathogenic variants from the neutral ones. The second theme of my work focuses on a machine learning model (“LYRUS”), which uses variation number as a feature, together with 14 other features, to predict the pathogenicity of human single amino acid variants. LYRUS was trained using a dataset that contained 22,639 single amino acid variants. The results indicated that LYRUS achieved comparable performance to current variant effect predictors. Predicting pathogenic variants is critical and understanding the pathogenic effects is also important. Compensated pathogenic deviations, which are alleles that are pathogenic in humans but neutral in some other species, are a special type of pathogenic variants. Studying compensated pathogenic deviations may help us to better understand the mechanism of certain genes. The third theme of my work focuses on comparative genomics analysis to identify compensated pathogenic deviations as well as the putative secondary sites that may mask the pathogenic effect of the pathogenic variants in ten breast cancer risk genes. Analysis of the possible compensatory variants revealed that they likely occur prior to the pathogenic deviations, and further research may assist to discover the underlying molecular mechanism for the compensating effect. In summary, my work utilized comparative genomic and machine learning approaches to predict the effect of human variants and identify variants that may mask the pathogenic effect of certain mutations.
- Notes:
- Thesis (Ph. D.)--Brown University, 2022
Citation
Lai, Jiaying,
"An Evolutionary Approach to Predict Human Pathogenic Mutations and Identify Putative Compensatory Sites"
(2022).
Center for Computational Molecular Biology Theses and Dissertations.
Brown Digital Repository. Brown University Library.
https://repository.library.brown.edu/studio/item/bdr:uqp2xx9u/
Relations
Collection:
-
Center for Computational Molecular Biology Theses and Dissertations
Theses and Dissertations for the Center for Computational Molecular Biology department....