From social to biological networks: a new algorithm uncovering key proteins in human disease

BEER-SHEVA, Israel, April 8th, 2025 – Researchers at Ben-Gurion University of the Negev have developed a machine-learning algorithm that could enhance our understanding of human biology and disease. The new method, Weighted Graph Anomalous Node Detection (WGAND), takes inspiration from social network analysis and is designed to identify proteins with significant roles in various human tissues.

WGAND was just published today in the Open Science journal GigaScience (https://doi.org/10.1093/gigascience/giaf034).

Proteins are essential molecules in our bodies, and they interact with each other in complex networks, known as protein-protein interaction (PPI) networks. Studying these networks helps scientists understand how proteins function and how they contribute to health and disease.

Prof. Esti Yeger-Lotem, Dr. Michael Fire, Dr. Jubran Juman, and Dr. Dima Kagan developed the algorithm to analyze these PPI networks to detect "anomalous" proteins—those that stand out due to their unique pattern of weighted interactions. This implies that the amount of the protein and its protein interactors is greater in that particular network, allowing them to carry more functions and drive more processes. This also indicates the great importance that these proteins have in the particular network, because the body will not waste energy on their production for no reason.

 The research combined the expertise in protein networks of Prof. Yeger-Lotem with the network analysis expertise of Dr. Fire derived from his study of social networks. In cybersecurity-related analyses of social networks, identifying atypical patterns can uncover fraudulent transactions or suspicious user behavior. The innovative insight of the new GigaScience study is that the same algorithms that uncover anomalies in social networks can be applied to the networks of proteins inside individual cells. By focusing on these anomalies, the algorithm can identify proteins that play crucial roles in specific tissues, such as the brain, heart, and liver.

WGAND successfully identified proteins associated with tissue-specific diseases, such as those involved in brain disorders and heart conditions. The algorithm also pinpointed proteins involved in critical biological processes, like neuron signalling in the brain and muscle contraction in the heart. Moreover, WGAND outperformed other existing methods in terms of accuracy and precision.

"This innovative algorithm has the potential to pinpoint which proteins are important in specific contexts, helping scientists to develop more targeted and effective treatments for various conditions," says Prof. Yeger-Lotem.

"It's exciting to see how bringing together expertise from bioinformatics and cybersecurity can lead to breakthroughs in understanding human biology. By applying network analysis and machine learning, we have developed a tool that helps uncover key proteins in different tissues—paving the way for new insights into human health and disease," says Dr. Michael Fire.

In accordance with Dr. Fire's principles as the head of the Fire AI Lab (https://fireailab.com), the WGAND algorithm is open source, allowing researchers worldwide to utilize and build upon it. The Yeger-Lotem lab also maintains web tools enabling easy access to researchers with no computational background (https://netbio.bgu.ac.il).

The research was supported by the Israel Science Foundation (Grant No. 401/22) and the Israeli Council for Higher Education (CHE) via the Data Science Research Center at Ben-Gurion University.

The authors Prof. Esti Yeger-Lotem and Dr. Michael Fire will be talking about their work in a free online webinar on April 8th at 3pm UTC and will be available to answer any questions. Sign up here:

https://cassyni.com/events/RfbzkzjvuLjxHYQb4AJdh7?yh30gdb

Read the Paper in GigaScience:

Dima Kagan, Juman Jubran, Esti Yeger-Lotem, and Michael Fire. Network-based anomaly detection algorithm reveals proteins with major roles in human tissues. GigaScience. 2025.

doi:10.1093/gigascience/giaf034