NGFN-PLUS
Bioinformatics analysis of Parkinson's disease
| Coordinator: | Prof. Dr. Andreas Zell | |
| Institution: | Center for Bioinformatics Tübingen, University of Tübingen | |
| Homepage: | www.cogsys.cs.uni-tuebingen.de |
Genome-wide association studies (GWAS) are a frequently used method to search for novel genetic risk factors or biomarkers. In this subproject, modern machine learning algorithms have been applied to augment traditional simple statistical approaches.
In a study that made use of four different GWAS data sets of Parkinson's disease we could show that is it generally possible to use those methods to make predictions about the individual disease risk, depending on the size of the data sets. For this, we developed the software MacLeaps, which automates such analyses of GWAS data sets without requiring detailed knowledge of the algorithms used. In an additional study we investigated and compared the predictive performance of various algorithms on data sets of additional diseases to achieve the best predictions possible. We also developed another approach and a software, the GWAS Pathway Identifier, which connects known biological knowledge from databases with the GWAS data to detect signalling and metabolic pathways that are potentially associated with the disease.
During the project, a model of a dopaminergic nerve cell was established with the purpose of identifying Parkinson relevant reaction fluxes in such a cell. The Systems Biology Markup Language (SBML) was used to define all relevant cell fluxes including the dopamine metabolism and transport, oxidative stress, aggregation of alpha-synuclein, lysosomal and proteasomal degradation, and mitophagy. For the investigation of the relevant fluxes, several experiments were performed using flux-balance-analysis to determine the strength of each reaction fluxes. The SBML model was published in the BioModels database with id MODEL1302200000.
Figure 1: Screenshot of MacLeaps that shows the results of a GWAS analysis.
In a study that made use of four different GWAS data sets of Parkinson's disease we could show that is it generally possible to use those methods to make predictions about the individual disease risk, depending on the size of the data sets. For this, we developed the software MacLeaps, which automates such analyses of GWAS data sets without requiring detailed knowledge of the algorithms used. In an additional study we investigated and compared the predictive performance of various algorithms on data sets of additional diseases to achieve the best predictions possible. We also developed another approach and a software, the GWAS Pathway Identifier, which connects known biological knowledge from databases with the GWAS data to detect signalling and metabolic pathways that are potentially associated with the disease.
During the project, a model of a dopaminergic nerve cell was established with the purpose of identifying Parkinson relevant reaction fluxes in such a cell. The Systems Biology Markup Language (SBML) was used to define all relevant cell fluxes including the dopamine metabolism and transport, oxidative stress, aggregation of alpha-synuclein, lysosomal and proteasomal degradation, and mitophagy. For the investigation of the relevant fluxes, several experiments were performed using flux-balance-analysis to determine the strength of each reaction fluxes. The SBML model was published in the BioModels database with id MODEL1302200000.
Figure 1: Screenshot of MacLeaps that shows the results of a GWAS analysis.
Figure 2: Schema of the dopaminergic nerve cell model, which consists of eleven sub-models.
KTT
MEDIA
CURRENT
NGFN-MEETING-2012
NGFN- MEETING
NGFN1&2
LINKS