Quantcast

Software IDs Molecular Interactions within Tumors

SpaceMarkers, a newly developed machine learning software by researchers at the Johns Hopkins Convergence Institute and the Johns Hopkins Kimmel Cancer Center, can identify molecular interactions among different types of cells in and around a tumor, researchers report..

According to the researchers, SpaceMarkers utilizes spatial transcriptomics, an advanced technology that enables the measurement of gene expression in tissue samples based on their cellular location. It is important to understand the molecular characteristics of individual cells and the effects of intercellular interactions in the tumor microenvironment (cells within and surrounding tumors) to gain insight into the factors contributing to tumor progression.

The introduction of SpaceMarkers and its applications in various cancer types was featured as the cover article in the April 2023 edition of the journal Cell Systems.

In many cases, highly expressed genes are associated with specific cell types that are overly abundant, dominant biological processes, or significant interactions among cell types not commonly found in healthy tissue. Dr. Elana Fertig, the senior author of the study and division director of oncology quantitative sciences and co-director of the Convergence Institute, explained that genes can be expressed in cells for various reasons, and certain genes can pose a high risk for cancer progression.

Dr. Fertig stated that SpaceMarkers can assist cancer researchers in determining whether a gene is overexpressed due to cell-to-cell interactions. It can also identify the specific cell-to-cell interactions associated with genes of interest. This new information can provide researchers with further avenues to understand the factors responsible for cancer or to address the question of why some patients respond to certain treatments while others do not.

According to Dr. Atul Deshpande, the lead author of the study and a postdoctoral researcher in the Fertig Lab at The Johns Hopkins University, SpaceMarkers operates by identifying regions of elevated activity from individual cell types as observed in spatial transcriptomic data. Regions exhibiting high activity from two cell types are recognized as sites of cell-to-cell interaction. Subsequently, the software algorithm detects molecular changes resulting from the interaction between these two cell types.

Dr. Deshpande explained that the software has two modes of operation. The first is a simple differential expression mode that identifies genes exhibiting significantly higher expression at sites of cell-to-cell interaction, suggesting that these interactions lead to increased gene expression rates. However, this mode does not take into account spatial variations in cell populations. The second mode (the residual mode) identifies genes with significantly higher expression after accounting for all cell populations identified in the spatial transcriptomic data.

The researchers tested SpaceMarkers using spatial transcriptomics data obtained from several clinical samples of pancreatic, breast, and liver cancers. The software’s validation was conducted by identifying genes known to influence interactions between tumor and immune cells.

Dr. Deshpande mentioned that many of the identified genes were consistent with the current understanding of tumor-immune interactions in these cancer types. Additionally, it was found that SpaceMarkers, in its current form, is better suited for the analysis of solid tumors, such as those found in some breast and liver cancers.

The research paper aimed to explore the capabilities of SpaceMarkers across different types of cancer, and in the next steps, the investigators plan to correlate the identified genes with treatments and patient responses. The authors intend to optimize SpaceMarkers in future studies to enhance its performance in cancers like pancreatic cancer, which lack a distinct tumor mass or well-defined boundaries. The Convergence Institute team also aims to expand the application of the software to analyze data from emerging high-resolution spatial technologies like multiplexed error-robust fluorescence in situ hybridization (MERFISH) or Xenium.

The study’s co-authors include Melanie Loth, Dimitrios Sidiropoulos, Shuming Zhang, Long Yuan, Alexander Bell, Qingffeng Zhu, Won Jin Ho, Cesar Santa-Maria, Daniele Gilkes, Stephen Williams, Cedric Uytingco, Jennifer Chew, Andrej Hartnett, Zachary Bent, Alexander Favorov, Aleksander Popel, Mark Yarchoan, Ashley Kiemen, Pei-Hsun Wu, Kohei Fujikura, Denis Wirtz, Laura Wood, Lei Zheng, Elizabeth Jaffee, Robert Anders, Ludmila Danilova, Genevieve Stein-O’Brien, and Luciane Kagohara were also listed as co-authors of the study.

In terms of potential conflicts of interest, it should be noted that Dr. Fertig serves on the Viosera Therapeutics scientific advisory board and works as a paid consultant for Merck and Mestag Therapeutics. These affiliations are managed in accordance with the conflict of interest policies of The Johns Hopkins University.




The material in this press release comes from the originating research organization. Content may be edited for style and length. Want more? Sign up for our daily email.