Quantifying tissue-specific effects on Parkinson's disease by Machine Learning
The way of DNA folds is changed by what is happening inside cells. The genes that are turned on, what is being repaired and replicated all change the DNA. Therefore, we can capture the shape of the DNA and use that information to work out how mutations cause disease. We will use machine learning to combine information on mutations, which genes are turned on in cells, and how the DNA is folded to work out how and where Parkinson's first develops. We have a spreadsheet that links Parkinson's mutations to the genes that they change. We will use this spreadsheet to read people's DNA sequence and convert it into a score that measures the risk of developing Parkinson's disease. We will be able to use this score to identify the parts of the body that change and cause Parkinson's disease. We will train our computer program using the UK Biobank data. We will then repeat the work using data from the Wellcome Trust Case and Control Consortium. To train our computer we need the individual genotype and medical data from the UK Biobank. Early tests have shown that this approach will make a big impact on our understanding of Parkinson's disease. Our results will improve our ability to predict a person's chances of developing Parkinson's disease. They will also identify new drug targets that may reduce Parkinson's disease risk in adults.