Prediction of depressive symptoms from socioeconomic data and DNA methylation signatures in depression

Depression is an increasingly common mental disorder associated with substantial deficits in the quality of life of the patient and increased mortality risk. Several Genomic-Wide Association Studies (GWAS) have been performed to identify genes associated with depression, but its partial heritability...

Descripción completa

Detalles Bibliográficos
Autor: Guerrero Simon, Laura
Tipo de recurso: tesis de maestría
Fecha de publicación:2020
País:España
Institución:Universitat Oberta de Catalunya (UOC)
Repositorio:O2, repositorio institucional de la UOC
OAI Identifier:oai:openaccess.uoc.edu:10609/122446
Acceso en línea:http://hdl.handle.net/10609/122446
Access Level:acceso abierto
Palabra clave:epigenetics
DNA methylation
depression
epigenética
metilación del ADN
depresión
epigenètica
metilació de l¿ADN
depressió
Bioinformatics -- TFM
Bioinformàtica -- TFM
Bioinformática -- TFM
Descripción
Sumario:Depression is an increasingly common mental disorder associated with substantial deficits in the quality of life of the patient and increased mortality risk. Several Genomic-Wide Association Studies (GWAS) have been performed to identify genes associated with depression, but its partial heritability among other characteristics suggests the involvement of epigenetic changes in the origin of the disease. Since has been proven a discrepancy between objective and subjective cognition in Major Depressive Disorder (MDD) patients, it is necessary to find a system to detect the disease from a biological perspective. Using a Canadian community-based cohort (n=94) containing DNA methylation data, stratified for early-life socioeconomic status, and assessed for depressive symptoms with the Center for Epidemiologic Studies Depression (CES-D) scale, a differential methylation analysis was performed. From this analysis, 31 cytosine guanine dinucleotides (CpG) were identified as differentially methylated in patients showing depressive symptoms from patients not showing those symptoms. The analysis was performed separating patients by gender and taking the variable age as a covariate. From the socioeconomic and biomolecular variables, and identified CpG sites, a random forest classifier was developed to create a depressive symptoms prediction tool. The resulting algorithm has an accuracy of 73.74% (repeated 15-fold cross-validation, with 3 repeats). The web application Desypre (http://desypre.000webhostapp.com/) was created to allow the public use of the classifier.