Artificial Intelligence (AI) has made its way to biological research with researchers revealing how it can be used in revealing patterns in huge amounts of gene expression data, and discover groups of genes related to disease. A paper on this new study has been published in Nature.
On social media platforms, we are used to getting friend suggestions. The platform does this on the basis of common friends. In this fashion, scientists attempted to build maps of biological networks based on how the different proteins and genes interact with each other. The researchers used artificial neural networks. These artificial networks are trained by experimental data, which means, in a network, programming is done that can bring results like those found in experiments. Now, when unknown data is fed into the trained network, it would analyze the data and decipher what it depicts. Artificial neural networks are excellent at finding patterns in enormous amount of complex data. For this, they are widely used in applications like image recognition. But this machine learning technique has seen very limited application in biological research.
“We have for the first time used deep learning to find disease-related genes. This is a very powerful method in the analysis of huge amounts of biological information, or ‘big data’,” said Sanjiv Dwivedi, a postdoctoral fellow in the Department of Physics, Chemistry and Biology (IFM) at Linköping University and first author of the published research.
The challenge associated is that it is not possible to see how exactly an artificial neural network solves a task. In a sense, it is a black box. He added, “We know what data we have fed into and the results, but the steps the network uses to bring the answer is not known. The researchers of the present study attempted to understand this.”
“When we analyzed our neural network, it turned out that the first hidden layer represented to a large extent interactions between various proteins. Deeper in the model, in contrast, on the third level, we found groups of different cell types. It’s extremely interesting that this type of biologically relevant grouping is automatically produced, given that our network has started from unclassified gene expression data,” said Mika Gustafsson, senior lecturer at IFM and corresponding author of the study.
Then, the researchers tried to find out whether their model of gene expression is successful at determining which gene expression patterns are associated with disease and which is normal. Their model turned out to be fruitful. The model finds relevant patterns that comply with the biological reality.
“We believe that the key to progress in the field is to understand the neural network. This can teach us new things about biological contexts, such as diseases in which many factors interact. And we believe that our method gives models that are easier to generalise and that can be used for many different types of biological information,” said Mika Gustafsson.
The researchers believe that successful use of AI to find gene patterns in diseases would pave the way for developing precision medicine in future.
