Scatterplot analysis of expression and DNA methylation integration data
Gene expression regulated by DNA methylation patterns has been long studied in relation to cancer. There exists a negative correlation between the expression of a gene and its methylation level. An integrative analysis of expression and methylation arrays was performed using three datasets for color...
| Autor: | |
|---|---|
| Formato: | tesis de maestría |
| Fecha de publicación: | 2018 |
| País: | España |
| Recursos: | Universitat Oberta de Catalunya (UOC) |
| Repositorio: | O2, repositorio institucional de la UOC |
| OAI Identifier: | oai:openaccess.uoc.edu:10609/82228 |
| Acesso em linha: | http://hdl.handle.net/10609/82228 |
| Access Level: | acceso abierto |
| Palavra-chave: | gene expression methylation integration expresión génica metilación integración expressió gènica metilació integració Bioinformatics -- TFM Bioinformàtica -- TFM Bioinformática -- TFM |
| Resumo: | Gene expression regulated by DNA methylation patterns has been long studied in relation to cancer. There exists a negative correlation between the expression of a gene and its methylation level. An integrative analysis of expression and methylation arrays was performed using three datasets for colorectal cancer: TCGA, GEO and own data. The datasets had over 11000 genes, 9000 of which were common. Based on the preconception that methylation represses expression, we selected genes that showed an L-shaped expression and methylation scatterplot with 4 different methods. The first method used, naive, was based on a signifi cant negative correlation. Another method was based on Conditional Mutual Information (CMI). A heuristic method was carried out by superimposing a grid on each scatterplot and weighing the cells according to an L-shape. Finally, a scagnostics selection analysis was based on 9 parameters de ning the shape of scatterplots. The scagnostics needed to be used in conjunction with other methods for optimal results. The accuracy, sensitivity and speci city were measured for the various methods and the one with the best diagnostic measures was the Heuristic, followed by the CMI and the naive. The one that fared lowest was the scagnostics. The final gene list was obtained from a pool of all methodologies.It resulted in 179 target genes, mostly coding for ATP-binding, transcription and zinc finger proteins. |
|---|