ImmunoTar—Integrative Prioritization of Cell Surface Targets for Cancer Immunotherapy

Artificial Intelligence Information Science Immunology Computer Science Oncology Life Sciences Medicine
cancer immunotherapy cell surface targets RNA sequencing proteomics computational tool

Cancer remains one of the leading causes of death globally. Despite significant advancements in immunotherapy in recent years, such as the successful application of chimeric antigen receptor T-cell (CAR-T) therapy and antibody-drug conjugates (ADCs), the effective identification of cancer-specific surface protein targets remains a major challenge in current research. The identification of surface protein targets is crucial for the development of precise and less toxic immunotherapies. Existing technologies, such as RNA sequencing and proteomics, can help researchers analyze these targets, but there is still a lack of systematic methods to prioritize the most suitable immunotherapy targets.

To address this issue, a research team from the Children’s Hospital of Philadelphia, Drexel University, BC Cancer Research Institute, and other institutions developed a computational tool called ImmunoTAR, which aims to systematically prioritize immunotherapy targets by integrating data from multiple public databases. This tool not only helps researchers more efficiently screen potential targets but also accelerates the development of novel immunotherapies. The research findings were published in Bioinformatics in 2025.

Research Background

Cancer immunotherapy has achieved significant success in recent years by activating the patient’s immune system to attack cancer cells. Particularly, CAR-T cell therapy and ADCs have demonstrated powerful efficacy in treating hematological malignancies and some solid tumors. However, the success of these therapies relies on the precise identification of cancer-specific surface proteins. Ideal immunotherapy targets should exhibit the following characteristics: high expression in cancer cells, low expression in normal tissues, clear surface localization, and functional relevance to tumors.

Existing technologies, such as RNA sequencing and mass spectrometry, can help researchers identify these targets, but due to the complexity and diversity of the data, how to systematically evaluate and prioritize these targets remains a challenge. To address this issue, the research team developed ImmunoTAR, which integrates user-provided RNA sequencing or proteomic data with quantitative features from multiple public databases to generate a score for each gene as an immunotherapy target.

Research Methods

Development and Workflow of ImmunoTAR

ImmunoTAR is a tool developed in the R programming language. Its main goal is to generate a score for each gene as an immunotherapy target by integrating user-provided cancer RNA sequencing or proteomic data with quantitative features from multiple public databases. The tool’s databases cover four major categories of data: normal tissue expression, protein localization, biological annotation, and reagent/therapeutic availability. Specifically, ImmunoTAR integrates data from GTEx (Genotype-Tissue Expression), EVO-DEVO (Mammalian Organ Development Project), CIRFESS (Compiled Interactive Resource for Extracellular and Surface Studies), Compartments, UniProt, DepMap, Gene Ontology (GO), Therapeutic Target Database (TTD), and others.

The workflow of ImmunoTAR consists of three main steps:

  1. Generating a Gene-Feature Data Matrix: Extracts summary features for each gene from the user-provided expression data and combines them with quantitative features from public databases to generate a gene-feature data matrix.
  2. Applying Project Analysis Parameters: Rescales and handles missing values in the data matrix, applies non-linear normalization (curving) and feature weights, and generates final feature values.
  3. Calculating Gene Scores: Computes the final score for each gene through weighted averaging and generates a table containing all feature values and gene scores.

Optimization and Validation

To optimize ImmunoTAR’s parameters, the research team used a proteomic dataset covering 12 pediatric cancer phenotypes and employed two strategies: multi-cancer optimization and phenotype-specific optimization. By comparing the scores from default parameters and optimized parameters, the team found that the multi-cancer optimization parameters performed more stably across multiple phenotypes, significantly improving the algorithm’s mean average precision score (MAP score).

To validate the effectiveness of ImmunoTAR, the research team tested it on multiple cancer datasets, including multiple myeloma (MM), Ewing sarcoma (EWS), and neuroblastoma (NBL). The results showed that ImmunoTAR not only effectively identified known immunotherapy targets but also discovered new potential targets.

Research Results

Multi-Cancer Optimization Parameters Significantly Improve Algorithm Performance

Through the multi-cancer optimization strategy, ImmunoTAR significantly increased the average MAP score by 27-fold across 12 pediatric cancer phenotypes. Specifically, the MAP scores for B-cell non-Hodgkin’s lymphoma and neuroblastoma reached 37% and 32%, respectively. Although phenotype-specific optimization parameters performed better in specific phenotypes, their performance in cross-phenotype applications was inferior to that of multi-cancer optimization parameters.

Application of ImmunoTAR in Multiple Myeloma

In the surface proteomics data of multiple myeloma (MM), ImmunoTAR successfully identified several known immunotherapy targets, such as ITGA4, ITGB7, and FLVCR1. These targets are closely related to cell adhesion, migration, and invasion in MM and exhibit limited expression in normal tissues. Additionally, ImmunoTAR identified several new potential targets, further validating its broad applicability across different cancer phenotypes.

Application of ImmunoTAR in Ewing Sarcoma

In the surface proteomics data of Ewing sarcoma (EWS), ImmunoTAR not only validated known targets such as ENPP1 and CDH11 but also discovered a new potential target, CADM1. CADM1 is expressed in various tumor types and is closely related to the biological functions of bone cancer. Although CADM1 is also expressed in some normal tissues, studies have shown that its toxicity as a target is low, indicating high potential for clinical application.

Application of ImmunoTAR in Neuroblastoma

In the surface proteomics data of neuroblastoma (NBL), ImmunoTAR successfully identified known targets such as L1CAM, ALK, NCAM1, and CD276, and discovered a new potential target, DLK1. DLK1 is highly expressed in NBL cells and is closely related to the undifferentiated state of tumor cells. A recent study also validated the potential of DLK1 as an immunotherapy target for NBL, further demonstrating the effectiveness of ImmunoTAR.

Discussion and Conclusion

As a tool for prioritizing cancer immunotherapy targets, ImmunoTAR provides a comprehensive and systematic method for target evaluation by integrating user-provided RNA sequencing and proteomic data with quantitative features from multiple public databases. Its validation across various cancer phenotypes demonstrates its ability to effectively identify known immunotherapy targets and discover new potential targets, showcasing its broad applicability.

Although ImmunoTAR excels in target prioritization, it still relies on high-quality data provided by users, and target validation requires further experimental research. Future studies could further expand ImmunoTAR’s functionality, such as integrating immunopeptidomics data and dual-target identification strategies, to enhance its application value in cancer immunotherapy.

The development of ImmunoTAR provides an efficient and systematic tool for the identification of cancer immunotherapy targets, potentially accelerating the development of novel immunotherapies and offering more treatment options for cancer patients.