There is currently no cure for neurodegenerative conditions; we urgently need new ways to understand them. In reality, any form of neurodegeneration is better considered several different conditions that look the same. Clinical trials could then test treatments targeted at the exact type of disease. We do not know the best way to subgroup each disease. One potential solution is to subgroup based on a combination of biological measures. Here I propose to use genetic profile, epigenetics (a system controlling whether genes are switched on or off), and the level of a nerve protein found in blood called neurofilament. I will test how much variations in genes, epigenetics and neurofilament change the risk of two conditions: amyotrophic lateral sclerosis (ALS), and frontotemporal dementia. I will then apply this to collections of samples with detailed measurements available from each person and use machine learning to find patterns that correspond to different subgroups. If I can confirm the subgroups, they can be used to group people for trials and to understand the underlying biology of the conditions.