News Release

New database maps proteins that could predict cancer

Reports and Proceedings

KTH, Royal Institute of Technology

Expression of protein in brain cancer

image: The yellow bar in this readout of protein GFAP indicates elevated expression in the blood of patients with brain tumors. (image: Human Disease Blood Atlas) view more 

Credit: Human Disease Blood Atlas

Cancer prediction medicine gained a boost with the recent unveiling of a new cancer protein profile database compiled from AI and machine learning. Announced by KTH Royal Institute of Technology Professor Mathias Uhlén, the new open-access Disease Blood Atlas provides a first-ever map of the proteome signature in blood from cancer patients.

The Disease Blood Atlas highlights 1,463 proteins associated with 12 different types of cancer, and presents proteins that can be used to identify individual cancer types based on a drop of blood.

The new database was produced by the Human Protein Atlas consortium, led by Uhlén and based within SciLifeLab, a joint research center that includes KTH Royal Institute of Technology, Uppsala University, Karolinska Institutet and Stockholm University.

Uhlén says that the Disease Blood Atlas was compiled from measurements of minute amounts of blood plasma collected from 1,400 cancer patients at the time of diagnosis and before treatment. The blood samples underwent a combination of statistical analysis of gene expression and machine-learning-based disease prediction.

“This is a novel pan-cancer strategy for exploring the proteome signature in blood from cancer patients,” he says.

The release marks the 22nd version of the open-access Human Protein Atlas, a resource for profiling human proteins (www.proteinatlas.org), which contains 12 sections each exploring the human proteins from different angels, including the new Disease Blood Atlas and the Protein 3-D Structure sections.

The release is accompanied by 5 million pages of updates in the Human Protein Atlas’ databases of tissues and cell lines.

The Protein 3-D Structure section shows the 3-D structures for all human proteins using an AI-based prediction model (AlfaFold). In addition, a major update of the Tissue Atlas section provides detailed multiplex spatial profiling of proteins specific for human testis and kidney. More data is also provided on single cell analysis of tissues and organs, as well as data from an extensive catalogue of human cell lines.

”We believe that the new sections of the open access Human Protein Atlas with large amounts of novel data covering all human proteins provides new dimensions of valuable information for researchers interested in human biology and disease,” Uhlén says.

The work was funded by the Knut and Alice Wallenberg Foundation

 


Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of news releases posted to EurekAlert! by contributing institutions or for the use of any information through the EurekAlert system.