����Vlog

When channelled effectively, technology has lifesaving benefits. One area where this impact is being felt is in the field of genomics. Mohammed Uddin, director of the Center for Applied and Translational Genomics and associate professor at the College of Medicine at Mohammed Bin Rashid University of Medicine and Health Sciences (MBRU) is optimistic about AI’s potential to advance genomics research and improve personalised medicine. MBRU is part of Dubai Health, which is an academic health system that connects several hospitals, medical centres and research institutions in Dubai. 

“We use multiple different technologies for genome sequencing, which largely focus on a technique known as long-read sequencing,” Uddin explains. His team has recently begun using a piece of automation software that takes variant files from long-read genome sequencing and interprets the data contained within them.

“To take the digital data being produced by these technologies, process and verify it would require a lot of manual work if we didn’t use AI or automation,” Uddin says. “For instance, a single genome might produce around 5 million genetic variants but only one of these may be causing a particular disease.”

“The use of automation software such as GenomeArc Horizon massively streamlines genomics research,” Uddin says. “Without it, lots of the analysis would have to be conducted manually.” Using the software, the team can carry out analysis on hundreds of genomes a day, whereas manually, it would take two to three days just to analyse one genome using guidelines developed by the American College of Medical Genetics and Genomics. If you’re looking at multiomics, which involves analysing data sets from multiple “omes”, including proteomes, metabolomes and more, in addition to genomes, automation can save you even more time, he adds.

 

AI works best when dealing with data that is accompanied by a pattern. It can identify this pattern to predict outcomes from data. “This is why AI is proving so useful to the field,” Uddin says. “Biology is full of patterns. For instance, we use another tool named Guppy, which identifies the base we are sequencing by looking at electrical signals. Each base has a similar pattern of signals. These signals are recognised by the tool’s AI, which makes our research much more efficient.”

To reach strong conclusions, genomic research requires data from large population sets – especially with rare genetic diseases. At the same time, personalised medicine requires a detailed picture of each individual. AI can be hugely helpful when working with large quantities of data.

“If you look at our direction of travel in the field, we need more and more data to make informed decisions, whether it is for diagnostics or drug development,” Uddin says. “Without automation, we aren’t making enough progress or we certainly aren’t making progress quickly enough. When it comes to personalised medicine, it simply wouldn’t be feasible to profile a whole community without automation.”

Automation and AI can not only speed up the delivery of genomic testing and personalised medicine but also significantly lower costs and reduce human error. However, there are also challenges to implementing AI in genomics. “An issue we’ve seen is that it can be difficult to isolate useful signals from all the noise around AI,” Uddin says. “We need tools that are validated, not simply hyped up. Another challenge is finding ways to train individuals to properly use AI tools. We are not there yet but more and more people are engaging with these tools and picking up relevant skills.” 

about the Mohammed Bin Rashid University of Medicine and Health Sciences.