19/02/2021

Artificial intelligence solves biology’s primary enigma: predicting the shape of proteins

AI is characterised as a tool capable of learning and analysing large amounts data related to patient history which, in turn, can help doctors offer better diagnoses and treatments

UIC Barcelona hosted the online conference “Artificial intelligence and medicine of the future”, given by a leading expert in personalised medicine and genomics. Dr Alfonso Valencia, ICREA research professor, director of the Department of Life Sciences at the Barcelona Supercomputing Centre (BSC-CNS), director of the Spanish National Bioinformatics Institute (INB-ISCIII) and director of the Spanish node of the European network ELIXIR, gave the conference organised by the Observatory on Artificial Intelligence and New Technologies (IOANT).

The event took place on Wednesday 17 February and was attended online by a one-hundred strong audience. It focused on how over recent years, advances in AI have meant that machines are now able to perform complex tasks quickly in the field of biomedicine.

“One of biology’s great unanswered questions has been almost completely answered long before we ever expected it would be”, said Alfonso Valencia, as he began his talk. For the first time, new and sophisticated AI methods, which have been applied to big data and knowledge, have found the keys to solving intricate protein structures. “This fact”, noted the bioinformatician, “is the most complex problem solved by artificial intelligence to date”.

Proteins are essential to life and their function depends largely on their three-dimensional structure, which is unique to each protein. So much so that discovering how this three-dimensional shape is determined has been one of biologists’ goals for years. On this basis, Alfonso Valencia explained how Google's DeepMind uses deep learning to predict how proteins acquire their form.

In this context, Valencia argued that “artificial intelligence will not replace doctors, but it will become a fierce ally”. Thanks to its rapid ability to learn and analyse large amounts data related to patient history, AI can therefore help doctors to offer their patients better diagnoses and treatments.

Watch the conference here for more details