Quantum Computing and Energy Efficiency
Using quantum physics to be energy efficient with High-Performance Computing simulations.
But first of all, let’s look at the science…
Haln researched endothelial cells (the cells lining the inside of the blood vessels) and wanted to better understand how they behave in pulmonary arterial hypertension (high blood pressure in the arteries around the lungs). “We know that endothelial cells are key players in initiating and driving the disease. They grow abnormally, become scarred and dysfunctional, and recruit inflammatory cells.” But instead of looking at endothelial cells as a whole, Haln wanted to see whether different endothelial cell subtypes respond to pulmonary hypertension in different ways.
To do so, Haln worked with mouse models of the disease, examining their endothelial lung cells and comparing these to endothelial lung cells of healthy mice. She used a method called single-cell RNA sequencing to look at the different levels of gene expression from the diseased tissue and healthy tissue. Haln explains that “single-cell RNA sequencing gives us information about the genes that are, say, activated or inactivated, in each cell”.
The results showed that there were different patterns of gene expression in the endothelial cells of diseased mice compared to healthy mice. This is specifically true for genes involved in the immune pathway. “We found a gene signature common to all endothelial subpopulations, involving the recruitment of T cells [a type of immune cell], and a gene signature unique to one subtype, involving blood vessel growth”.
To confirm this finding, Haln and her colleagues (Julie & Sweta Sweta) analysed bulk RNA-sequencing data from additional mice, and from other species, including rats and humans. They found that 51% of the upregulated mouse genes were also upregulated in rats and humans. Bulk RNA sequencing was particularly useful to strengthen the research findings so far, especially if species had fewer endothelial cells to work with. They used Eddie for the bulk-RNA sequencing, the University of Edinburgh’s supercomputer.
More about Eddie
Expression levels differ based on the location of the endothelial cell across the blood vessel.
Back to the web-based application…
Having briefly outlined the study, it becomes obvious that mapping played a major role throughout the project. To name a few examples, there was mapping of gene expression levels, mapping of different endothelial subtypes and mapping of different blood vessel locations. To help with visualising these outputs, Haln made an R Shiny app in R Studio whilst also working on her manuscript. She wanted to publish the R Shiny app on the web, so researchers could explore genes of interest in the app independently.
“I knew R Shiny had its own sever, but we were looking for something cheaper”. This is how Haln discovered Eleanor, the University’s cloud computing tool. Eleanor lets you create virtual machines configured to your exact requirements. You can build anything that you could with a physical computer.
More about Eleanor“Eleanor was cheaper, and had other advantages too: we had support at hand, and we knew that the servers were safely located with the University of Edinburgh. Having better control and better support was important to us, especially if we want to work with human data in the future, which requires more secure environments”.
This was the first time Haln tried to transfer something like this onto the web. There was a rough guideline from her department, but she got most help from the Research Services Wiki and from Mike Wallis at the Eleanor drop-in sessions.
View Drop-in SessionsHer biggest challenge was estimating how big the app was going to be, i.e. how much CPU and RAM was needed. Eleanor is free up to a certain volume, beyond which you have to pay a fee – but only for the amount you actually use. Mike Wallis was helpful in estimating the size and the cost, and also in finding ways to reduce the size of the project to further reduce the cost.
All in all, getting the web application up and running took about two weeks. Haln’s top tip: “Get in touch with Research Services, they are incredibly helpful. Even if somebody else in our department had done something similar, I still found engaging with Research Services incredibly useful as each project is different and it gave you access to technical expertise that we needed.”
Screenshot of web-based application - researchers can enter any genes they are interested in and see how it is expressed along the artery-capillary-vein axis.
View applicationThis case study was written by Dr Sarah Janac, CMVM Research Facilitator. Many thanks for contributions and comments from Dr Haln Chen and Dr Julie Rodor.