| Summary: | 碩士 === 國立成功大學 === 資訊工程學系 === 104 === Synthetic lethality (SL) is an interaction between two genes, which means the cell will be alive if one of these two genes is disabled but dead if both genes are disabled. SL was discovered in 1946 but become more popular recently because there are some cancer therapies based on this interaction. Due to this special interaction, we can make cancer cells lethal but normal cells alive by targeting the synthetic lethal pair of the mutant gene in cancer. However, it costs a lot to find an SL using traditional experiments. Therefore, we propose a new SL prediction system to easily predict some potential SL in silico.
In this study, we design an SL prediction system based on a text-mining method and an inference model. First, we extend some essential genes using the text-mining method. Compared with mutant genes, there are only a few essential genes in a few cancers recorded in databases. With these potential essential genes extended using the text-mining method, we can predict more SL not restricted by scant essential genes. Second, we predict SL from mutant genes and extend essential genes and filter wrong gene pairs using gene co-expression and the co-occurrence of two genes in the literature.
We also compare our prediction system with experimental screening data. Recently, there has been a lot of SL discovered through a few screening experiments, but it is hard for biological researchers to find more important SL in these screening data. Through our prediction system, biological researchers can find potential SL more easily. We also study some cases of colon cancer and find some information that reveals some novel potential SL might be valid.
This research predicts potential SL by combining a text-mining method and gene data. We extract more essential genes from literature mining to complement the lack of essential gene data. We then predict some potential SL using an inference model and compare the results with screening data to allow biological researchers find interesting SL quickly.
|
|---|
