Project Title: Using machine learning to predict cell-type specific effects of genetic variants which influence genome regulation
Supervisor: Dr Nathan Skene
Location: Sir Michael Uren Hub, White City Campus
As a child, I decided early on to not work in medicine; blood, guts and screaming did not suit me. Instead, like, I imagine, many engineering students, I began my time at university wanting – quite literally – to do rocket science. This conviction gradually died down; I discovered, and soon realised I preferred, the intersection between mathematics, programming and biology – what I started to consider one of the true cutting-edge fields of our time.
After finishing my BSc at KTH in Sweden, I decided to move abroad for my master's studies. During my time at Imperial College London, my preference for this gratifyingly blood-less biology has grown into a keen interest in the applications for mathematics and machine-learning in medicine. My master's thesis consisted of a project in which I analysed time-lapse microscopy videos of cancer cells marked with mRuby-PCNA to infer the time of entry and exit into and out of different cell cycle stages.
Upon graduating from Imperial College London, I was offered a job as a Data Management Consultant at Echo State in Stockholm, Sweden. My main project was planning and implementing a cloud-based data lake in a data science team at a large industrial company. I gained a lot of experience working with data and programming in a professional setting. Still, I soon realised that I thrive the most when I continuously push my knowledge boundaries in an academic environment.
I am now undertaking a PhD where I am using deep learning to predict how DNA sequences changes alter genomic regulatory features under Dr Nathan Skene's supervision.
- 2021-Present: PhD Clinical Medicine Science (Brain Science) at Imperial College London, UK
- 2019-2021: Data Management Consultant at Echo State, Sweden
- 2018-2019: MSc Applied Mathematics at Imperial College London, UK
- 2015-2018: BSc Engineering Physics at KTH - Royal Institute of Technology, Sweden
The development of new technologies for analysing biological samples, e.g. next-generation sequencing, has caused a massive increase in the amount of data generated from biological experiments. One consequence of this is that deep learning - a technique that has proven massively successful when applied to image analysis or natural language processing - can be used to analyse genetic data. I am researching how new advancements in deep learning research can be used in predicting how changes in the DNA sequence affect the regulation of gene expression. The goal is to be able to understand the functional role of genetic variants that are associated with neurodegenerative diseases and to identify potential drug targets.