News

Cracking cancer’s code through functional connections

Anna Crysler
July 2, 2025

Among cancer researchers, their computers, servers and databanks store thousands of terabytes of omics data, enabling novel discoveries about genetic and proteomic relationships. However, making meaningful connections can be computationally challenging. What if there were a way to harness the power of machine learning to help interpret this data and identify unrecognized patterns that advance therapeutic strategies?

A recent paper in Nature Cancer introduces a new tool for decoding and uncovering functional connections within cancer biology. FunMap, a machine-learning-driven platform, allows researchers to understand how genes and proteins work together in cancer, even when they aren’t directly connected. Bing Zhang, a professor of molecular and human genetics at Baylor College of Medicine, and his lab aim to bridge the gap between large-scale cancer omics data and functional interpretation using machine learning.

The team used large-scale proteogenomic data, or integrated information about genes, RNA and proteins, across 11 cancer types to chart a functional network of more than 10,000 genes. Unlike traditional protein–protein interaction networks, which focus on physical contacts between proteins, FunMap assesses “cofunctionality,” the concept that genes or proteins can participate in the same biological process even if they do not physically interact.

“Think of a complex research lab,” Zhiao Shi, lead programmer in the lab and first author of the paper, explained the computational tool. “A computational biologist and a wet lab scientist may never perform experiments together, but the computational analysis is crucial for guiding the wet lab experiments and interpreting results. Though they do not interact directly, their roles are tightly coordinated to achieve scientific breakthroughs — this is cofunction.”

With the ability to incorporate graph-neural-network-based deep learning, a type of model that learns from data structured as networks of connected elements, FunMap can identify cancer driver mutations with low frequencies. This expands the understanding of cancer pathogenesis beyond high-frequency mutations and may potentiate new discoveries in cancer diagnostics and treatment.

FunMap also advances functional genomics by shedding light on understudied cancer genes, such as RBM34 and MAB21L4, also known as dark genes. These understudied genes and their protein counterparts have not been studied in the context of cancer but are significantly over or under expressed in tumors. Shi explained that their approach “enables a more systematic and data-driven assignment of functions to poorly characterized cancer-associated genes, aiding in the discovery of novel cancer biology.”

The platform is available to the public at funmap.linkedomics.org, where scientists can explore the network and apply it to their own studies. The Zhang lab plans to expand its tool with additional data types, such as epigenomics and protein modification.

“By identifying key cancer-associated proteins and functional pathways, our findings can help prioritize therapeutic targets, ultimately contributing to the development of more effective treatments,” Shi said. “In the long run, this research could lead to improved cancer diagnostics and therapies, benefiting patients by making precision medicine more actionable and impactful.”

Overview of FunMap, a machine learning and network-based framework that integrates pan-cancer proteogenomic data to identify functional modules, predict understudied protein functions, and discover low-frequency cancer drivers.
Bing Zhang, Baylor College of Medicine
Overview of FunMap, a machine learning and network-based framework that integrates pan-cancer proteogenomic data to identify functional modules, predict understudied protein functions, and discover low-frequency cancer drivers.

Enjoy reading ASBMB Today?

Become a member to receive the print edition four times a year and the digital edition monthly.

Learn more
Anna Crysler

Anna Crysler holds a B.A. in biochemistry from Albion College and is a is a Ph.D. student in bioengineering at the University of Pennsylvania. She is an ASBMB Today volunteer contributor.

Get the latest from ASBMB Today

Enter your email address, and we’ll send you a weekly email with recent articles, interviews and more.

Latest in Science

Science highlights or most popular articles

Hope for a cure hangs on research
Essay

Hope for a cure hangs on research

July 17, 2025

Amid drastic proposed cuts to biomedical research, rare disease families like Hailey Adkisson’s fight for survival and hope. Without funding, science can’t “catch up” to help the patients who need it most.

Before we’ve lost what we can’t rebuild: Hope for prion disease
Feature

Before we’ve lost what we can’t rebuild: Hope for prion disease

July 15, 2025

Sonia Vallabh and Eric Minikel, a husband-and-wife team racing to cure prion disease, helped develop ION717, an antisense oligonucleotide treatment now in clinical trials. Their mission is personal — and just getting started.

Defeating deletions and duplications
News

Defeating deletions and duplications

July 11, 2025

Promising therapeutics for chromosome 15 rare neurodevelopmental disorders, including Angelman syndrome, Dup15q syndrome and Prader–Willi syndrome.

Using 'nature’s mistakes' as a window into Lafora disease
Feature

Using 'nature’s mistakes' as a window into Lafora disease

July 10, 2025

After years of heartbreak, Lafora disease families are fueling glycogen storage research breakthroughs, helping develop therapies that may treat not only Lafora but other related neurological disorders.

Gaze into the proteomics crystal ball
In-person Conference

Gaze into the proteomics crystal ball

July 1, 2025

The 15th International Symposium on Proteomics in the Life Sciences symposium will be held August 17–21 in Cambridge, Massachusetts.

Bacterial enzyme catalyzes body odor compound formation
Journal News

Bacterial enzyme catalyzes body odor compound formation

June 27, 2025

Researchers identify a skin-resident Staphylococcus hominis dipeptidase involved in creating sulfur-containing secretions. Read more about this recent Journal of Biological Chemistry paper.