In the realm of biology, understanding the intricate functions of a cell remains a monumental goal. The fusion of vast data quantities and the rapid evolution of artificial intelligence (AI) methodologies empower scientists to make significant strides toward this objective. A dedicated team of researchers is at the forefront of decoding the molecular language of cells, viewing them as minuscule yet complex factories teeming with molecular components, with proteins acting as the essential workers.
The challenge of protein research
Despite recent advancements in protein research, where the structure of approximately 200,000 individual proteins and many larger molecular complexes have been identified, full comprehension of all cellular functions is still out of reach. With its 20,000 genes, the human genome produces a myriad of protein forms through various modifications. These proteins, made up of amino acids, fold into intricate three-dimensional structures, the total variety and critical forms of which remain to be fully understood.
The quest for a complete protein map
Professor Arne Elofsson of Stockholm University is leading this exciting venture, based at SciLifeLab and supported by the Knut and Alice Wallenberg Foundation. Elofsson’s goal is to map most protein interactions within a human cell, requiring the development of new AI-based methods and insights from large-scale experiments.
AI’s revolutionary impact
The field of AI has significantly accelerated this study. In 2020, DeepMind’s AlphaFold algorithm, which is capable of accurately predicting the structure of individual proteins, marked a turning point. Following its release to the research community in 2021, it has fueled the publication of several hundred scientific papers. Elofsson highlights AlphaFold’s transformative impact on studying protein structures, thanks to its deep learning foundations and training on all known proteins.
The current project is an interdisciplinary collaboration involving experts from Stockholm University, KTH Royal Institute of Technology, and Uppsala University. Their expertise spans bioinformatics, computer science, graph neural networks, and mass spectrometry. The team plans to utilize novel machine-learning methods to identify various proteins (protoforms) and study their interactions using AI.
Verification techniques
The researchers will verify their findings through various methods. Cross-linking will be employed to examine protein binding to specific DNA sequences. Mass spectrometry will play a crucial role in determining the composition of protein complexes detecting proteins even in minute concentrations.
Despite international competition, Elofsson remains optimistic about Sweden’s standing, bolstered by significant investments in data-driven life sciences and access to Berzelius, Sweden’s fastest AI and machine learning supercomputer. He emphasizes the need for developing universally applicable methods to advance research.
This research is poised to offer a comprehensive understanding of the molecular components of a human cell and their interactions, marking a significant milestone in biology. It will enhance our knowledge of complex biological processes and diseases at the molecular level, paving the way for simulations of entire cells and studies of their functions at unimaginable detail levels.
